diff --git a/README.md b/README.md
index ad8305d..3ba722d 100644
--- a/README.md
+++ b/README.md
@@ -141,14 +141,14 @@ ctest --test-dir build --output-on-failure --verbose
 - [x] **Phase 2**: In-memory query execution (SeqScan, Filter, Projection)
 - [x] **Phase 3**: Disk-based storage with buffer pool
 - [x] **Phase 4**: BTree indexes with query planner integration
-- [ ] **Phase 5**: JOIN operations
+- [x] **Phase 5**: JOIN operations
   - [x] Parse `INNER JOIN ... ON ...` with qualified column references
   - [x] Execute joins via `NestedLoopJoin`
   - [x] Add rule-based join algorithm choice (`NestedLoopJoin` vs `HashJoin`)
   - [x] Support `JOIN + WHERE` (single-table pushdown + post-join filter)
   - [x] Add correctness checks (ambiguous columns, swapped `ON` sides, type-mismatch safety)
-  - [ ] Add `EXPLAIN` command in REPL to print physical plan (`SeqScan`/`IndexScan`/`Join` path)
-  - [ ] Add join-condition index matching (index-aware join access path, e.g. index nested-loop opportunities)
+  - [x] Add `EXPLAIN` command in REPL to print physical plan (`SeqScan`/`IndexScan`/`Join` path)
+  - [x] Add join-condition index matching (`IndexNestedLoopJoin` when index exists on join column)
 - [ ] **Phase 6**: Transactions
 
 ## Architecture
diff --git a/src/execution/CMakeLists.txt b/src/execution/CMakeLists.txt
index 5e595b6..421cfde 100644
--- a/src/execution/CMakeLists.txt
+++ b/src/execution/CMakeLists.txt
@@ -6,6 +6,7 @@ add_library(execution
     index_scan.cpp
     nested_loop_join.cpp
     hash_join.cpp
+    index_nested_loop_join.cpp
 )
 
 target_link_libraries(execution PUBLIC optimizer storage common)
diff --git a/src/execution/executor.cpp b/src/execution/executor.cpp
index b92fb99..f93ba00 100644
--- a/src/execution/executor.cpp
+++ b/src/execution/executor.cpp
@@ -258,6 +258,8 @@ namespace sql
             return ExecuteCreateIndex(static_cast<CreateIndexStatement *>(stmt));
         case StatementType::DROP_TABLE:
             return ExecuteDropTable(static_cast<DropTableStatement *>(stmt));
+        case StatementType::EXPLAIN_STMT:
+            return ExecuteExplain(static_cast<ExplainStatement *>(stmt));
         default:
             return {false, "Unsupported statement type", {}, {}};
         }
@@ -388,6 +390,36 @@ namespace sql
             EnsureJoinContextTable(left, right);
             return std::make_unique<HashJoin>(left, right, left_col, right_col, node->join_build_right);
         }
+        case PhysicalPlanType::INDEX_NESTED_LOOP_JOIN:
+        {
+            Table *left = catalog_->GetTable(node->table_name);
+            Table *right = catalog_->GetTable(node->right_table_name);
+            if (left == nullptr || right == nullptr)
+                throw std::runtime_error("JOIN table not found while building operator tree");
+
+            // Resolve which ON-clause column belongs to which table (handles swapped ON order).
+            const auto [left_col, right_col] = ResolveJoinColumns(node, left, right);
+
+            // join_right_as_outer: true → right is outer, left is inner (has index)
+            //                      false → left is outer, right is inner (has index)
+            const bool right_is_outer = node->join_right_as_outer;
+            Table *outer_table = right_is_outer ? right : left;
+            Table *inner_table = right_is_outer ? left : right;
+            const std::string outer_col = right_is_outer ? right_col : left_col;
+            const std::string inner_col = right_is_outer ? left_col : right_col;
+            const std::string &inner_table_name = right_is_outer ? node->table_name : node->right_table_name;
+
+            BTree *inner_index = catalog_->GetIndex(inner_table_name, inner_col);
+            if (inner_index == nullptr)
+                throw std::runtime_error("Expected index not found on " + inner_table_name + "." + inner_col);
+
+            EnsureJoinContextTable(left, right);
+            // outer_is_left: left table is outer when right_is_outer=false
+            const bool outer_is_left = !right_is_outer;
+            return std::make_unique<IndexNestedLoopJoin>(
+                outer_table, inner_table, inner_index,
+                outer_col, inner_col, outer_is_left);
+        }
         case PhysicalPlanType::FILTER:
         {
             if (node->children.empty())
@@ -408,7 +440,8 @@ namespace sql
             auto child = BuildOperatorTree(node->children[0].get(), table, join_table);
             Table *projection_join_table = join_table;
             if (node->children[0]->type == PhysicalPlanType::NESTED_LOOP_JOIN ||
-                node->children[0]->type == PhysicalPlanType::HASH_JOIN)
+                node->children[0]->type == PhysicalPlanType::HASH_JOIN ||
+                node->children[0]->type == PhysicalPlanType::INDEX_NESTED_LOOP_JOIN)
             {
                 projection_join_table = catalog_->GetTable(node->children[0]->right_table_name);
             }
@@ -692,4 +725,24 @@ namespace sql
         return {true, "Table '" + drop->table + "' dropped.", {}, {}};
     }
 
+    ExecutionResult Executor::ExecuteExplain(ExplainStatement *explain)
+    {
+        ExecutionResult result;
+        try
+        {
+            materialized_tables_.clear();
+            join_context_table_.reset();
+            Optimizer optimizer;
+            auto physical_plan = optimizer.BuildPhysicalPlan(explain->select.get(), catalog_);
+            result.success = true;
+            result.message = optimizer.ExplainPhysicalPlan(physical_plan.get());
+        }
+        catch (const std::exception &e)
+        {
+            result.success = false;
+            result.message = e.what();
+        }
+        return result;
+    }
+
 } // namespace sql
diff --git a/src/execution/executor.h b/src/execution/executor.h
index eb70d14..039633e 100644
--- a/src/execution/executor.h
+++ b/src/execution/executor.h
@@ -7,6 +7,7 @@
 #include "execution/index_scan.h"
 #include "execution/nested_loop_join.h"
 #include "execution/hash_join.h"
+#include "execution/index_nested_loop_join.h"
 #include "parser/ast.h"
 #include "optimizer/optimizer.h"
 #include "catalog/catalog.h"
@@ -44,6 +45,7 @@ namespace sql
         ExecutionResult ExecuteUpdate(UpdateStatement *update);
         ExecutionResult ExecuteCreateIndex(CreateIndexStatement *create);
         ExecutionResult ExecuteDropTable(DropTableStatement *drop);
+        ExecutionResult ExecuteExplain(ExplainStatement *explain);
 
         // Evaluate an expression (reused for INSERT values, UPDATE SET, etc.)
         Value EvaluateExpr(const Expression *expr, const Tuple *tuple = nullptr, Table *table = nullptr) const;
diff --git a/src/execution/index_nested_loop_join.cpp b/src/execution/index_nested_loop_join.cpp
new file mode 100644
index 0000000..e8b271a
--- /dev/null
+++ b/src/execution/index_nested_loop_join.cpp
@@ -0,0 +1,76 @@
+#include "execution/index_nested_loop_join.h"
+#include <stdexcept>
+
+namespace sql
+{
+
+    namespace
+    {
+        std::string StripQualifier(const std::string &name)
+        {
+            size_t dot = name.find('.');
+            if (dot == std::string::npos)
+                return name;
+            return name.substr(dot + 1);
+        }
+    } // namespace
+
+    void IndexNestedLoopJoin::Open()
+    {
+        if (outer_table_ == nullptr || inner_table_ == nullptr || inner_index_ == nullptr)
+            throw std::runtime_error("IndexNestedLoopJoin: null table or index");
+
+        outer_col_idx_ = outer_table_->GetColumnIndex(StripQualifier(outer_col_));
+        if (outer_col_idx_ < 0)
+            throw std::runtime_error("IndexNestedLoopJoin: unknown outer column: " + outer_col_);
+
+        outer_cursor_ = 0;
+        inner_matches_.clear();
+        inner_cursor_ = 0;
+    }
+
+    bool IndexNestedLoopJoin::Next(Tuple *tuple)
+    {
+        const auto &outer_rows = outer_table_->GetTuples();
+
+        while (true)
+        {
+            // Consume remaining inner matches for current outer row
+            while (inner_cursor_ < inner_matches_.size())
+            {
+                size_t inner_row_idx = inner_matches_[inner_cursor_++];
+                const Tuple &inner_row = inner_table_->GetTuple(inner_row_idx);
+
+                std::vector<Value> joined;
+                const Tuple &left_row = outer_is_left_ ? current_outer_ : inner_row;
+                const Tuple &right_row = outer_is_left_ ? inner_row : current_outer_;
+                joined.reserve(left_row.GetValueCount() + right_row.GetValueCount());
+                for (size_t i = 0; i < left_row.GetValueCount(); ++i)
+                    joined.push_back(left_row.GetValue(i));
+                for (size_t i = 0; i < right_row.GetValueCount(); ++i)
+                    joined.push_back(right_row.GetValue(i));
+
+                *tuple = Tuple(std::move(joined));
+                return true;
+            }
+
+            // Advance to next outer row
+            if (outer_cursor_ >= outer_rows.size())
+                return false;
+
+            current_outer_ = outer_rows[outer_cursor_++];
+            Value probe_key = current_outer_.GetValue(static_cast<size_t>(outer_col_idx_));
+            inner_matches_ = inner_index_->Search(probe_key);
+            inner_cursor_ = 0;
+        }
+    }
+
+    void IndexNestedLoopJoin::Close()
+    {
+        outer_cursor_ = 0;
+        inner_matches_.clear();
+        inner_cursor_ = 0;
+        outer_col_idx_ = -1;
+    }
+
+} // namespace sql
diff --git a/src/execution/index_nested_loop_join.h b/src/execution/index_nested_loop_join.h
new file mode 100644
index 0000000..bfcc57c
--- /dev/null
+++ b/src/execution/index_nested_loop_join.h
@@ -0,0 +1,52 @@
+#pragma once
+
+#include "execution/operator.h"
+#include "storage/table.h"
+#include "storage/btree.h"
+#include <string>
+#include <vector>
+
+namespace sql
+{
+
+    // Index Nested-Loop Join: for each outer row, probes inner table via BTree index.
+    // Output column order is always (left_table_cols..., right_table_cols...) regardless
+    // of which side is outer; outer_is_left controls concatenation order.
+    class IndexNestedLoopJoin : public Operator
+    {
+    public:
+        // outer_table:   table iterated row by row
+        // inner_table:   table probed via index
+        // inner_index:   BTree on inner_table.inner_col
+        // outer_col:     column name in outer_table used as probe key
+        // inner_col:     indexed column name in inner_table
+        // outer_is_left: if true, output is (outer || inner); else (inner || outer)
+        IndexNestedLoopJoin(Table *outer_table, Table *inner_table,
+                            BTree *inner_index,
+                            std::string outer_col, std::string inner_col,
+                            bool outer_is_left)
+            : outer_table_(outer_table), inner_table_(inner_table),
+              inner_index_(inner_index),
+              outer_col_(std::move(outer_col)), inner_col_(std::move(inner_col)),
+              outer_is_left_(outer_is_left) {}
+
+        void Open() override;
+        bool Next(Tuple *tuple) override;
+        void Close() override;
+
+    private:
+        Table *outer_table_;
+        Table *inner_table_;
+        BTree *inner_index_;
+        std::string outer_col_;
+        std::string inner_col_;
+        bool outer_is_left_;
+
+        int outer_col_idx_ = -1;
+        size_t outer_cursor_ = 0;
+        Tuple current_outer_;
+        std::vector<size_t> inner_matches_;
+        size_t inner_cursor_ = 0;
+    };
+
+} // namespace sql
diff --git a/src/lexer/lexer.cpp b/src/lexer/lexer.cpp
index 3227d2a..03b4878 100644
--- a/src/lexer/lexer.cpp
+++ b/src/lexer/lexer.cpp
@@ -268,6 +268,8 @@ namespace sql
              {"NOT", TokenType::NOT},
              {"DROP", TokenType::DROP},
              {"JOIN", TokenType::JOIN},
+             {"INNER", TokenType::INNER},
+             {"EXPLAIN", TokenType::EXPLAIN},
              {"TRUE", TokenType::TRUE},
              {"FALSE", TokenType::FALSE}};
 
diff --git a/src/lexer/token.h b/src/lexer/token.h
index 43af3da..765b306 100644
--- a/src/lexer/token.h
+++ b/src/lexer/token.h
@@ -33,6 +33,8 @@ namespace sql
         NOT,
         DROP,
         JOIN,
+        EXPLAIN,
+        INNER,
 
         // Identifiers and Literals
         IDENTIFIER,
diff --git a/src/optimizer/optimizer.cpp b/src/optimizer/optimizer.cpp
index 11fc2c6..0075142 100644
--- a/src/optimizer/optimizer.cpp
+++ b/src/optimizer/optimizer.cpp
@@ -386,22 +386,69 @@ namespace sql
             const size_t left_count = table->GetTupleCount();
             const size_t right_count = right_table->GetTupleCount();
 
-            // Rule-based join choice:
-            // - HASH_JOIN for larger equi-joins
-            // - NESTED_LOOP_JOIN for small inputs
-            const size_t total_rows = left_count + right_count;
-            const bool use_hash_join = total_rows >= 16;
-
-            auto join = std::make_unique<PhysicalPlanNode>(
-                use_hash_join ? PhysicalPlanType::HASH_JOIN : PhysicalPlanType::NESTED_LOOP_JOIN);
+            // Resolve which join column belongs to which table.
+            // The ON clause may have columns in either order (e.g., ON right.x = left.y),
+            // so we must check before looking up indexes.
+            const auto left_side = ResolveColumnSide(*select->join_left_column,
+                                                      select->table, *select->join_table,
+                                                      table, right_table);
+            const auto right_side = ResolveColumnSide(*select->join_right_column,
+                                                       select->table, *select->join_table,
+                                                       table, right_table);
+
+            // Determine the actual column name for each table side.
+            std::string col_on_left_table;  // column belonging to select->table
+            std::string col_on_right_table; // column belonging to select->join_table
+            if (left_side == PredicateTableSide::LEFT && right_side == PredicateTableSide::RIGHT)
+            {
+                col_on_left_table = StripQualifier(*select->join_left_column);
+                col_on_right_table = StripQualifier(*select->join_right_column);
+            }
+            else if (left_side == PredicateTableSide::RIGHT && right_side == PredicateTableSide::LEFT)
+            {
+                col_on_left_table = StripQualifier(*select->join_right_column);
+                col_on_right_table = StripQualifier(*select->join_left_column);
+            }
+            // else: ambiguous or unresolved — skip INLJ
+
+            // Check if either join column has an index → prefer INDEX_NESTED_LOOP_JOIN.
+            std::unique_ptr<PhysicalPlanNode> join;
+            if (!col_on_left_table.empty() && !col_on_right_table.empty())
+            {
+                BTree *left_index = catalog->GetIndex(select->table, col_on_left_table);
+                BTree *right_index = catalog->GetIndex(*select->join_table, col_on_right_table);
+
+                if (right_index != nullptr)
+                {
+                    // Right table has index → right is inner, left is outer.
+                    join = std::make_unique<PhysicalPlanNode>(PhysicalPlanType::INDEX_NESTED_LOOP_JOIN);
+                    join->join_right_as_outer = false; // left is outer
+                }
+                else if (left_index != nullptr)
+                {
+                    // Left table has index → left is inner, right is outer.
+                    join = std::make_unique<PhysicalPlanNode>(PhysicalPlanType::INDEX_NESTED_LOOP_JOIN);
+                    join->join_right_as_outer = true; // right is outer
+                }
+            }
+
+            if (join == nullptr)
+            {
+                // Rule-based choice between HASH_JOIN and NESTED_LOOP_JOIN.
+                const size_t total_rows = left_count + right_count;
+                const bool use_hash_join = total_rows >= 16;
+                join = std::make_unique<PhysicalPlanNode>(
+                    use_hash_join ? PhysicalPlanType::HASH_JOIN : PhysicalPlanType::NESTED_LOOP_JOIN);
+                // Iterate smaller table in outer loop for nested loop.
+                join->join_right_as_outer = right_count < left_count;
+                // Build hash table on smaller side for hash join.
+                join->join_build_right = right_count <= left_count;
+            }
+
             join->table_name = select->table;
             join->right_table_name = *select->join_table;
             join->join_left_column = *select->join_left_column;
             join->join_right_column = *select->join_right_column;
-            // Rule-based choice: iterate smaller table in outer loop for nested loop.
-            join->join_right_as_outer = right_count < left_count;
-            // Rule-based choice: build hash table on smaller side for hash join.
-            join->join_build_right = right_count <= left_count;
             join->children.push_back(std::move(left_input));
             join->children.push_back(std::move(right_input));
             current = std::move(join);
@@ -484,6 +531,19 @@ namespace sql
                 << ", build=" << (node->join_build_right ? "right" : "left")
                 << ")";
             break;
+        case PhysicalPlanType::INDEX_NESTED_LOOP_JOIN:
+        {
+            const std::string &outer = node->join_right_as_outer ? node->right_table_name : node->table_name;
+            const std::string &inner = node->join_right_as_outer ? node->table_name : node->right_table_name;
+            const std::string &inner_col = node->join_right_as_outer ? node->join_left_column : node->join_right_column;
+            out << pad << "IndexNestedLoopJoin(outer=" << outer
+                << ", inner=" << inner
+                << "(index on " << inner_col << ")"
+                << ", on=" << node->join_left_column
+                << " = " << node->join_right_column
+                << ")";
+            break;
+        }
         case PhysicalPlanType::FILTER:
             out << pad << "Filter";
             if (node->predicate != nullptr)
diff --git a/src/optimizer/optimizer.h b/src/optimizer/optimizer.h
index af87504..cc3f3e2 100644
--- a/src/optimizer/optimizer.h
+++ b/src/optimizer/optimizer.h
@@ -41,6 +41,7 @@ namespace sql
         INDEX_SCAN,
         NESTED_LOOP_JOIN,
         HASH_JOIN,
+        INDEX_NESTED_LOOP_JOIN,
         FILTER,
         PROJECTION
     };
@@ -65,10 +66,10 @@ namespace sql
         std::optional<Value> high_key;
         bool high_inclusive = true;
 
-        // NESTED_LOOP_JOIN
+        // NESTED_LOOP_JOIN / HASH_JOIN / INDEX_NESTED_LOOP_JOIN
         std::string join_left_column;
         std::string join_right_column;
-        bool join_right_as_outer = false;
+        bool join_right_as_outer = false; // NLJ: right is outer loop; INLJ: right is outer (left has index)
 
         // HASH_JOIN
         bool join_build_right = true;
diff --git a/src/parser/ast.h b/src/parser/ast.h
index 5f9fc32..0defab1 100644
--- a/src/parser/ast.h
+++ b/src/parser/ast.h
@@ -59,7 +59,8 @@ namespace sql
         INSERT,
         CREATE_INDEX,
         DELETE_STMT,
-        UPDATE_STMT
+        UPDATE_STMT,
+        EXPLAIN_STMT
     };
 
     struct Statement
@@ -130,6 +131,12 @@ namespace sql
         StatementType GetType() const override { return StatementType::UPDATE_STMT; }
     };
 
+    struct ExplainStatement : public Statement
+    {
+        std::unique_ptr<SelectStatement> select;
+        StatementType GetType() const override { return StatementType::EXPLAIN_STMT; }
+    };
+
     std::string ExpressionTypeToString(ExpressionType type);
     std::string StatementTypeToString(StatementType type);
     std::string DumpExpression(const Expression *expr, int indent = 0);
diff --git a/src/parser/parser.cpp b/src/parser/parser.cpp
index 721d067..64f776a 100644
--- a/src/parser/parser.cpp
+++ b/src/parser/parser.cpp
@@ -53,6 +53,8 @@ namespace sql
             return ParseUpdate();
         case TokenType::DROP:
             return ParseDropTable();
+        case TokenType::EXPLAIN:
+            return ParseExplain();
         default:
             throw std::runtime_error("Unexpected token at start of statement: " + current_token_.value);
         }
@@ -79,7 +81,18 @@ namespace sql
         Token table = Expect(TokenType::IDENTIFIER);
         stmt->table = table.value;
 
-        if (Match(TokenType::JOIN))
+        bool has_join = false;
+        if (Match(TokenType::INNER))
+        {
+            Expect(TokenType::JOIN);
+            has_join = true;
+        }
+        else if (Match(TokenType::JOIN))
+        {
+            has_join = true;
+        }
+
+        if (has_join)
         {
             Token join_table = Expect(TokenType::IDENTIFIER);
             stmt->join_table = join_table.value;
@@ -363,4 +376,12 @@ namespace sql
         return name;
     }
 
+    std::unique_ptr<ExplainStatement> Parser::ParseExplain()
+    {
+        Expect(TokenType::EXPLAIN);
+        auto stmt = std::make_unique<ExplainStatement>();
+        stmt->select = ParseSelect();
+        return stmt;
+    }
+
 } // namespace sql
diff --git a/src/parser/parser.h b/src/parser/parser.h
index aa9e491..e43fe8b 100644
--- a/src/parser/parser.h
+++ b/src/parser/parser.h
@@ -22,6 +22,7 @@ namespace sql
 
         std::unique_ptr<Statement> ParseCreate();
         std::unique_ptr<SelectStatement> ParseSelect();
+        std::unique_ptr<ExplainStatement> ParseExplain();
         std::unique_ptr<CreateTableStatement> ParseCreateTable();
         std::unique_ptr<CreateIndexStatement> ParseCreateIndex();
         std::unique_ptr<DropTableStatement> ParseDropTable();
diff --git a/test/integration/query_test.cpp b/test/integration/query_test.cpp
index c8398f4..6e8b470 100644
--- a/test/integration/query_test.cpp
+++ b/test/integration/query_test.cpp
@@ -462,4 +462,101 @@ namespace sql
         EXPECT_EQ(result.tuples[0].GetValue(1).GetAsInt(), 99);
     }
 
+    // ── EXPLAIN ───────────────────────────────────────────────────────────────────
+
+    TEST(IntegrationTest, ExplainSeqScan)
+    {
+        Catalog catalog;
+        RunSQL(catalog, "CREATE TABLE t (id INTEGER, name VARCHAR(50));");
+        auto result = RunSQL(catalog, "EXPLAIN SELECT * FROM t;");
+        EXPECT_TRUE(result.success);
+        EXPECT_NE(result.message.find("SeqScan"), std::string::npos);
+        EXPECT_NE(result.message.find("Projection"), std::string::npos);
+    }
+
+    TEST(IntegrationTest, ExplainIndexScan)
+    {
+        Catalog catalog;
+        RunSQL(catalog, "CREATE TABLE t (id INTEGER);");
+        RunSQL(catalog, "INSERT INTO t VALUES (1), (2), (3);");
+        RunSQL(catalog, "CREATE INDEX idx_id ON t (id);");
+        auto result = RunSQL(catalog, "EXPLAIN SELECT * FROM t WHERE id = 2;");
+        EXPECT_TRUE(result.success);
+        EXPECT_NE(result.message.find("IndexScan"), std::string::npos);
+    }
+
+    TEST(IntegrationTest, ExplainJoin)
+    {
+        Catalog catalog;
+        RunSQL(catalog, "CREATE TABLE orders (id INTEGER, cid INTEGER);");
+        RunSQL(catalog, "CREATE TABLE customers (id INTEGER, name VARCHAR(50));");
+        auto result = RunSQL(catalog, "EXPLAIN SELECT * FROM orders JOIN customers ON orders.cid = customers.id;");
+        EXPECT_TRUE(result.success);
+        // Without index: should be hash or nested loop join
+        EXPECT_TRUE(result.message.find("Join") != std::string::npos);
+    }
+
+    // ── INDEX NESTED-LOOP JOIN ────────────────────────────────────────────────────
+
+    TEST(IntegrationTest, IndexNestedLoopJoinBasic)
+    {
+        Catalog catalog;
+        RunSQL(catalog, "CREATE TABLE orders (oid INTEGER, cid INTEGER);");
+        RunSQL(catalog, "CREATE TABLE customers (id INTEGER, name VARCHAR(50));");
+        RunSQL(catalog, "INSERT INTO orders VALUES (1, 10), (2, 20), (3, 10);");
+        RunSQL(catalog, "INSERT INTO customers VALUES (10, 'Alice'), (20, 'Bob');");
+        RunSQL(catalog, "CREATE INDEX idx_cid ON customers (id);");
+
+        auto result = RunSQL(catalog, "SELECT * FROM orders JOIN customers ON orders.cid = customers.id;");
+        EXPECT_TRUE(result.success);
+        ASSERT_EQ(result.tuples.size(), 3);
+    }
+
+    TEST(IntegrationTest, IndexNestedLoopJoinExplain)
+    {
+        Catalog catalog;
+        RunSQL(catalog, "CREATE TABLE orders (oid INTEGER, cid INTEGER);");
+        RunSQL(catalog, "CREATE TABLE customers (id INTEGER, name VARCHAR(50));");
+        RunSQL(catalog, "INSERT INTO orders VALUES (1, 10);");
+        RunSQL(catalog, "INSERT INTO customers VALUES (10, 'Alice');");
+        RunSQL(catalog, "CREATE INDEX idx_cid ON customers (id);");
+
+        auto explain = RunSQL(catalog, "EXPLAIN SELECT * FROM orders JOIN customers ON orders.cid = customers.id;");
+        EXPECT_TRUE(explain.success);
+        EXPECT_NE(explain.message.find("IndexNestedLoopJoin"), std::string::npos);
+    }
+
+    TEST(IntegrationTest, IndexNestedLoopJoinSwappedOnColumns)
+    {
+        Catalog catalog;
+        RunSQL(catalog, "CREATE TABLE orders (oid INTEGER, cid INTEGER);");
+        RunSQL(catalog, "CREATE TABLE customers (id INTEGER, name VARCHAR(50));");
+        RunSQL(catalog, "INSERT INTO orders VALUES (1, 10), (2, 20), (3, 10);");
+        RunSQL(catalog, "INSERT INTO customers VALUES (10, 'Alice'), (20, 'Bob');");
+        RunSQL(catalog, "CREATE INDEX idx_cid ON customers (id);");
+
+        // ON clause has columns in reverse order: right table column first.
+        auto result = RunSQL(catalog, "SELECT * FROM orders JOIN customers ON customers.id = orders.cid;");
+        EXPECT_TRUE(result.success);
+        ASSERT_EQ(result.tuples.size(), 3);
+
+        // EXPLAIN should still pick IndexNestedLoopJoin despite swapped ON order.
+        auto explain = RunSQL(catalog, "EXPLAIN SELECT * FROM orders JOIN customers ON customers.id = orders.cid;");
+        EXPECT_TRUE(explain.success);
+        EXPECT_NE(explain.message.find("IndexNestedLoopJoin"), std::string::npos);
+    }
+
+    TEST(IntegrationTest, InnerJoinSyntax)
+    {
+        Catalog catalog;
+        RunSQL(catalog, "CREATE TABLE a (id INTEGER, val INTEGER);");
+        RunSQL(catalog, "CREATE TABLE b (aid INTEGER, name VARCHAR(50));");
+        RunSQL(catalog, "INSERT INTO a VALUES (1, 100), (2, 200);");
+        RunSQL(catalog, "INSERT INTO b VALUES (1, 'x'), (2, 'y');");
+
+        auto result = RunSQL(catalog, "SELECT * FROM a INNER JOIN b ON a.id = b.aid;");
+        EXPECT_TRUE(result.success);
+        ASSERT_EQ(result.tuples.size(), 2);
+    }
+
 } // namespace sql