diff --git a/pkg/sql/parser/performance_regression_norace.go b/pkg/sql/parser/performance_regression_norace.go
index 7eb3ee7a..8ec4a064 100644
--- a/pkg/sql/parser/performance_regression_norace.go
+++ b/pkg/sql/parser/performance_regression_norace.go
@@ -1,5 +1,4 @@
 //go:build !race
-// +build !race
 
 package parser
 
diff --git a/pkg/sql/parser/performance_regression_race.go b/pkg/sql/parser/performance_regression_race.go
index f53c8545..8de2c614 100644
--- a/pkg/sql/parser/performance_regression_race.go
+++ b/pkg/sql/parser/performance_regression_race.go
@@ -1,5 +1,4 @@
 //go:build race
-// +build race
 
 package parser
 
diff --git a/pkg/sql/parser/recovery.go b/pkg/sql/parser/recovery.go
new file mode 100644
index 00000000..7eb76202
--- /dev/null
+++ b/pkg/sql/parser/recovery.go
@@ -0,0 +1,122 @@
+package parser
+
+import (
+	"fmt"
+
+	"github.com/ajitpratap0/GoSQLX/pkg/models"
+	"github.com/ajitpratap0/GoSQLX/pkg/sql/ast"
+	"github.com/ajitpratap0/GoSQLX/pkg/sql/token"
+)
+
+// ParseError represents a parse error with position information.
+type ParseError struct {
+	Msg       string
+	TokenIdx  int
+	Line      int
+	Column    int
+	TokenType string
+	Literal   string
+}
+
+func (e *ParseError) Error() string {
+	if e.Line > 0 {
+		return fmt.Sprintf("parse error at line %d, column %d (token %d): %s", e.Line, e.Column, e.TokenIdx, e.Msg)
+	}
+	return fmt.Sprintf("parse error at token %d: %s", e.TokenIdx, e.Msg)
+}
+
+// isStatementStartingKeyword checks if the current token is a statement-starting keyword.
+func (p *Parser) isStatementStartingKeyword() bool {
+	if p.currentToken.ModelType != modelTypeUnset {
+		switch p.currentToken.ModelType {
+		case models.TokenTypeSelect, models.TokenTypeInsert, models.TokenTypeUpdate,
+			models.TokenTypeDelete, models.TokenTypeCreate, models.TokenTypeAlter,
+			models.TokenTypeDrop, models.TokenTypeWith, models.TokenTypeMerge,
+			models.TokenTypeRefresh, models.TokenTypeTruncate:
+			return true
+		}
+	}
+	// Fallback: string comparison for tokens without ModelType (e.g., tests)
+	switch string(p.currentToken.Type) {
+	case "SELECT", "INSERT", "UPDATE", "DELETE", "CREATE", "ALTER", "DROP",
+		"WITH", "MERGE", "REFRESH", "TRUNCATE":
+		return true
+	}
+	return false
+}
+
+// synchronize advances the parser past the current error to a synchronization point:
+// either past a semicolon or to a statement-starting keyword.
+func (p *Parser) synchronize() {
+	for p.currentPos < len(p.tokens) && !p.isType(models.TokenTypeEOF) {
+		// If we hit a semicolon, consume it and stop
+		if p.isType(models.TokenTypeSemicolon) {
+			p.advance()
+			return
+		}
+		// If we hit a statement-starting keyword, stop (don't consume it)
+		if p.isStatementStartingKeyword() {
+			return
+		}
+		p.advance()
+	}
+}
+
+// ParseWithRecovery parses a token stream, recovering from errors to collect multiple
+// errors and return a partial AST with successfully parsed statements.
+//
+// Unlike Parse(), which stops at the first error, this method uses synchronization
+// tokens (semicolons and statement-starting keywords) to skip past errors and
+// continue parsing subsequent statements.
+//
+// Parameters:
+//   - tokens: Slice of parser tokens to parse
+//
+// Returns:
+//   - []ast.Statement: Successfully parsed statements (may be empty)
+//   - []error: All parse errors encountered (each includes position information)
+func (p *Parser) ParseWithRecovery(tokens []token.Token) ([]ast.Statement, []error) {
+	p.tokens = tokens
+	p.currentPos = 0
+	if len(tokens) > 0 {
+		p.currentToken = tokens[0]
+	}
+
+	var statements []ast.Statement
+	var errors []error
+
+	for p.currentPos < len(tokens) && !p.isType(models.TokenTypeEOF) {
+		// Skip semicolons between statements
+		if p.isType(models.TokenTypeSemicolon) {
+			p.advance()
+			continue
+		}
+
+		savedPos := p.currentPos
+		stmt, err := p.parseStatement()
+		if err != nil {
+			// Create a ParseError with position info
+			loc := p.currentLocation()
+			pe := &ParseError{
+				Msg:      err.Error(),
+				TokenIdx: savedPos,
+				Line:     loc.Line,
+				Column:   loc.Column,
+			}
+			if savedPos < len(tokens) {
+				pe.TokenType = string(tokens[savedPos].Type)
+				pe.Literal = tokens[savedPos].Literal
+			}
+			errors = append(errors, pe)
+			p.synchronize()
+		} else {
+			statements = append(statements, stmt)
+			// Optionally consume semicolon after statement
+			if p.isType(models.TokenTypeSemicolon) {
+				p.advance()
+			}
+		}
+	}
+
+	return statements, errors
+}
diff --git a/pkg/sql/parser/recovery_multi_error_test.go b/pkg/sql/parser/recovery_multi_error_test.go
new file mode 100644
index 00000000..085b3946
--- /dev/null
+++ b/pkg/sql/parser/recovery_multi_error_test.go
@@ -0,0 +1,177 @@
+package parser
+
+import (
+	"testing"
+
+	"github.com/ajitpratap0/GoSQLX/pkg/sql/token"
+)
+
+func eof() token.Token {
+	return token.Token{Type: "EOF", Literal: ""}
+}
+
+func semi() token.Token {
+	return token.Token{Type: ";", Literal: ";"}
+}
+
+func tok(typ, lit string) token.Token {
+	return token.Token{Type: token.Type(typ), Literal: lit}
+}
+
+// TestParseWithRecovery_MultipleErrors tests that multiple syntax errors are all reported.
+func TestParseWithRecovery_MultipleErrors(t *testing.T) {
+	// "INVALID1 foo; INVALID2 bar;"
+	tokens := []token.Token{
+		tok("IDENT", "INVALID1"), tok("IDENT", "foo"), semi(),
+		tok("IDENT", "INVALID2"), tok("IDENT", "bar"), semi(),
+		eof(),
+	}
+	p := NewParser()
+	stmts, errs := p.ParseWithRecovery(tokens)
+	if len(stmts) != 0 {
+		t.Errorf("expected 0 statements, got %d", len(stmts))
+	}
+	if len(errs) < 2 {
+		t.Errorf("expected at least 2 errors, got %d", len(errs))
+	}
+	// Each error should be a *ParseError with position info
+	for i, err := range errs {
+		if _, ok := err.(*ParseError); !ok {
+			t.Errorf("error %d is not a *ParseError: %T", i, err)
+		}
+	}
+}
+
+// TestParseWithRecovery_FirstValidSecondInvalid tests partial AST with valid+invalid mix.
+func TestParseWithRecovery_FirstValidSecondInvalid(t *testing.T) {
+	// "SELECT * FROM users; INVALID foo;"
+	tokens := []token.Token{
+		tok("SELECT", "SELECT"), tok("*", "*"), tok("FROM", "FROM"), tok("IDENT", "users"), semi(),
+		tok("IDENT", "INVALID"), tok("IDENT", "foo"), semi(),
+		eof(),
+	}
+	p := NewParser()
+	stmts, errs := p.ParseWithRecovery(tokens)
+	if len(stmts) != 1 {
+		t.Errorf("expected 1 statement, got %d", len(stmts))
+	}
+	if len(errs) != 1 {
+		t.Errorf("expected 1 error, got %d", len(errs))
+	}
+}
+
+// TestParseWithRecovery_AllInvalid tests that all-invalid input returns empty AST + multiple errors.
+func TestParseWithRecovery_AllInvalid(t *testing.T) {
+	tokens := []token.Token{
+		tok("IDENT", "BAD1"), semi(),
+		tok("IDENT", "BAD2"), semi(),
+		tok("IDENT", "BAD3"), semi(),
+		eof(),
+	}
+	p := NewParser()
+	stmts, errs := p.ParseWithRecovery(tokens)
+	if len(stmts) != 0 {
+		t.Errorf("expected 0 statements, got %d", len(stmts))
+	}
+	if len(errs) != 3 {
+		t.Errorf("expected 3 errors, got %d", len(errs))
+	}
+}
+
+// TestParseWithRecovery_UnclosedParen tests recovery after unclosed parenthesis.
+func TestParseWithRecovery_UnclosedParen(t *testing.T) {
+	// "SELECT (1 + ; SELECT * FROM users;"
+	tokens := []token.Token{
+		tok("SELECT", "SELECT"), tok("(", "("), tok("INT", "1"), tok("+", "+"), semi(),
+		tok("SELECT", "SELECT"), tok("*", "*"), tok("FROM", "FROM"), tok("IDENT", "users"), semi(),
+		eof(),
+	}
+	p := NewParser()
+	stmts, errs := p.ParseWithRecovery(tokens)
+	if len(errs) < 1 {
+		t.Errorf("expected at least 1 error, got %d", len(errs))
+	}
+	if len(stmts) < 1 {
+		t.Errorf("expected at least 1 successfully parsed statement, got %d", len(stmts))
+	}
+}
+
+// TestParseWithRecovery_InvalidExpression tests recovery after invalid expression.
+func TestParseWithRecovery_InvalidExpression(t *testing.T) {
+	// "SELECT FROM; SELECT 1;"
+	// First SELECT has no columns (invalid), second is valid
+	tokens := []token.Token{
+		tok("SELECT", "SELECT"), tok("FROM", "FROM"), semi(),
+		tok("SELECT", "SELECT"), tok("INT", "1"), semi(),
+		eof(),
+	}
+	p := NewParser()
+	stmts, errs := p.ParseWithRecovery(tokens)
+	// The first SELECT FROM might parse differently depending on parser internals,
+	// but we should get at least one statement or error
+	if len(stmts)+len(errs) < 2 {
+		t.Errorf("expected at least 2 total statements+errors, got stmts=%d errs=%d", len(stmts), len(errs))
+	}
+}
+
+// TestParseWithRecovery_RecoveryToKeyword tests recovery skipping to next statement keyword.
+func TestParseWithRecovery_RecoveryToKeyword(t *testing.T) {
+	// "INVALID foo bar SELECT * FROM users;"
+	// No semicolon after invalid part, should recover at SELECT keyword
+	tokens := []token.Token{
+		tok("IDENT", "INVALID"), tok("IDENT", "foo"), tok("IDENT", "bar"),
+		tok("SELECT", "SELECT"), tok("*", "*"), tok("FROM", "FROM"), tok("IDENT", "users"), semi(),
+		eof(),
+	}
+	p := NewParser()
+	stmts, errs := p.ParseWithRecovery(tokens)
+	if len(errs) != 1 {
+		t.Errorf("expected 1 error, got %d", len(errs))
+	}
+	if len(stmts) != 1 {
+		t.Errorf("expected 1 statement, got %d", len(stmts))
+	}
+}
+
+// TestParseWithRecovery_EmptyInput tests empty token stream.
+func TestParseWithRecovery_EmptyInput(t *testing.T) {
+	tokens := []token.Token{eof()}
+	p := NewParser()
+	stmts, errs := p.ParseWithRecovery(tokens)
+	if len(stmts) != 0 {
+		t.Errorf("expected 0 statements, got %d", len(stmts))
+	}
+	if len(errs) != 0 {
+		t.Errorf("expected 0 errors, got %d", len(errs))
+	}
+}
+
+// TestParseWithRecovery_AllValid tests that all-valid input returns all statements.
+func TestParseWithRecovery_AllValid(t *testing.T) {
+	tokens := []token.Token{
+		tok("SELECT", "SELECT"), tok("INT", "1"), semi(),
+		tok("SELECT", "SELECT"), tok("INT", "2"), semi(),
+		eof(),
+	}
+	p := NewParser()
+	stmts, errs := p.ParseWithRecovery(tokens)
+	if len(stmts) != 2 {
+		t.Errorf("expected 2 statements, got %d", len(stmts))
+	}
+	if len(errs) != 0 {
+		t.Errorf("expected 0 errors, got %d", len(errs))
+	}
+}
+
+// TestParseError_ErrorMessage tests ParseError formatting.
+func TestParseError_ErrorMessage(t *testing.T) {
+	e := &ParseError{Msg: "unexpected token", TokenIdx: 5}
+	if e.Error() != "parse error at token 5: unexpected token" {
+		t.Errorf("unexpected error message: %s", e.Error())
+	}
+
+	e2 := &ParseError{Msg: "bad syntax", TokenIdx: 3, Line: 2, Column: 10}
+	if e2.Error() != "parse error at line 2, column 10 (token 3): bad syntax" {
+		t.Errorf("unexpected error message: %s", e2.Error())
+	}
+}
diff --git a/pkg/sql/tokenizer/norace.go b/pkg/sql/tokenizer/norace.go
index 03f8f581..94e3829a 100644
--- a/pkg/sql/tokenizer/norace.go
+++ b/pkg/sql/tokenizer/norace.go
@@ -1,5 +1,4 @@
 //go:build !race
-// +build !race
 
 package tokenizer
 
diff --git a/pkg/sql/tokenizer/race.go b/pkg/sql/tokenizer/race.go
index 99b1ff7b..f1aae592 100644
--- a/pkg/sql/tokenizer/race.go
+++ b/pkg/sql/tokenizer/race.go
@@ -1,5 +1,4 @@
 //go:build race
-// +build race
 
 package tokenizer