diff --git a/flow/connectors/bigquery/merge_stmt_generator.go b/flow/connectors/bigquery/merge_stmt_generator.go
index eb8ebb6177..59a269a092 100644
--- a/flow/connectors/bigquery/merge_stmt_generator.go
+++ b/flow/connectors/bigquery/merge_stmt_generator.go
@@ -100,6 +100,44 @@ func (m *mergeStmtGenerator) generateFlattenedCTE() string {
 		m.syncBatchID, m.dstTableName)
 }
 
+// This function is to support datatypes like JSON which cannot be partitioned by or compared by BigQuery
+func (m *mergeStmtGenerator) transformedPkeyStrings(forPartition bool) []string {
+	pkeys := make([]string, 0, len(m.normalizedTableSchema.PrimaryKeyColumns))
+	columnNameTypeMap := make(map[string]qvalue.QValueKind, len(m.normalizedTableSchema.Columns))
+	for _, col := range m.normalizedTableSchema.Columns {
+		columnNameTypeMap[col.Name] = qvalue.QValueKind(col.Type)
+	}
+
+	for _, pkeyCol := range m.normalizedTableSchema.PrimaryKeyColumns {
+		pkeyColType, ok := columnNameTypeMap[pkeyCol]
+		if !ok {
+			continue
+		}
+		switch pkeyColType {
+		case qvalue.QValueKindJSON:
+			if forPartition {
+				pkeys = append(pkeys, fmt.Sprintf("TO_JSON_STRING(%s)", m.shortColumn[pkeyCol]))
+			} else {
+				pkeys = append(pkeys, fmt.Sprintf("TO_JSON_STRING(_t.`%s`)=TO_JSON_STRING(_d.%s)",
+					pkeyCol, m.shortColumn[pkeyCol]))
+			}
+		case qvalue.QValueKindFloat32, qvalue.QValueKindFloat64:
+			if forPartition {
+				pkeys = append(pkeys, fmt.Sprintf("CAST(%s as STRING)", m.shortColumn[pkeyCol]))
+			} else {
+				pkeys = append(pkeys, fmt.Sprintf("_t.`%s`=_d.%s", pkeyCol, m.shortColumn[pkeyCol]))
+			}
+		default:
+			if forPartition {
+				pkeys = append(pkeys, m.shortColumn[pkeyCol])
+			} else {
+				pkeys = append(pkeys, fmt.Sprintf("_t.`%s`=_d.%s", pkeyCol, m.shortColumn[pkeyCol]))
+			}
+		}
+	}
+	return pkeys
+}
+
 // generateDeDupedCTE generates a de-duped CTE.
 func (m *mergeStmtGenerator) generateDeDupedCTE() string {
 	const cte = `_dd AS (
@@ -111,13 +149,8 @@ func (m *mergeStmtGenerator) generateDeDupedCTE() string {
 			WHERE _peerdb_rank=1
 	) SELECT * FROM _dd`
 
-	shortPkeys := make([]string, 0, len(m.normalizedTableSchema.PrimaryKeyColumns))
-	for _, pkeyCol := range m.normalizedTableSchema.PrimaryKeyColumns {
-		shortPkeys = append(shortPkeys, m.shortColumn[pkeyCol])
-	}
-
-	pkeyColsStr := fmt.Sprintf("(CONCAT(%s))", strings.Join(shortPkeys,
-		", '_peerdb_concat_', "))
+	shortPkeys := m.transformedPkeyStrings(true)
+	pkeyColsStr := strings.Join(shortPkeys, ",")
 	return fmt.Sprintf(cte, pkeyColsStr)
 }
 
@@ -151,11 +184,7 @@ func (m *mergeStmtGenerator) generateMergeStmt(unchangedToastColumns []string) s
 	}
 	updateStringToastCols := strings.Join(updateStatementsforToastCols, " ")
 
-	pkeySelectSQLArray := make([]string, 0, len(m.normalizedTableSchema.PrimaryKeyColumns))
-	for _, pkeyColName := range m.normalizedTableSchema.PrimaryKeyColumns {
-		pkeySelectSQLArray = append(pkeySelectSQLArray, fmt.Sprintf("_t.%s=_d.%s",
-			pkeyColName, m.shortColumn[pkeyColName]))
-	}
+	pkeySelectSQLArray := m.transformedPkeyStrings(false)
 	// t.<pkey1> = d.<pkey1> AND t.<pkey2> = d.<pkey2> ...
 	pkeySelectSQL := strings.Join(pkeySelectSQLArray, " AND ")
 
diff --git a/flow/e2e/bigquery/peer_flow_bq_test.go b/flow/e2e/bigquery/peer_flow_bq_test.go
index 9d0b8f6a57..74a1949ff7 100644
--- a/flow/e2e/bigquery/peer_flow_bq_test.go
+++ b/flow/e2e/bigquery/peer_flow_bq_test.go
@@ -1514,3 +1514,56 @@ func (s PeerFlowE2ETestSuiteBQ) Test_Soft_Delete_Insert_After_Delete() {
 	require.NoError(s.t, err)
 	require.Equal(s.t, int64(0), numNewRows)
 }
+
+func (s PeerFlowE2ETestSuiteBQ) Test_JSON_PKey_BQ() {
+	env := e2e.NewTemporalTestWorkflowEnvironment(s.t)
+
+	srcTableName := s.attachSchemaSuffix("test_json_pkey_bq")
+	dstTableName := "test_json_pkey_bq"
+
+	_, err := s.Conn().Exec(context.Background(), fmt.Sprintf(`
+	CREATE TABLE IF NOT EXISTS %s (
+		id SERIAL NOT NULL,
+		j JSON NOT NULL,
+		key TEXT NOT NULL,
+		value TEXT NOT NULL
+	);
+	`, srcTableName))
+	require.NoError(s.t, err)
+
+	_, err = s.Conn().Exec(context.Background(), fmt.Sprintf(`
+		ALTER TABLE %s REPLICA IDENTITY FULL
+		`, srcTableName))
+	require.NoError(s.t, err)
+
+	connectionGen := e2e.FlowConnectionGenerationConfig{
+		FlowJobName:      s.attachSuffix("test_json_pkey_flow"),
+		TableNameMapping: map[string]string{srcTableName: dstTableName},
+		Destination:      s.bqHelper.Peer,
+		CdcStagingPath:   "",
+	}
+
+	flowConnConfig := connectionGen.GenerateFlowConnectionConfigs()
+	flowConnConfig.MaxBatchSize = 100
+
+	go func() {
+		e2e.SetupCDCFlowStatusQuery(s.t, env, connectionGen)
+		// insert 10 rows into the source table
+		for i := range 10 {
+			testKey := fmt.Sprintf("test_key_%d", i)
+			testValue := fmt.Sprintf("test_value_%d", i)
+			testJson := `'{"name":"jack", "age":12, "spouse":null}'::json`
+			_, err = s.Conn().Exec(context.Background(), fmt.Sprintf(`
+			INSERT INTO %s(key, value, j) VALUES ($1, $2, %s)
+		`, srcTableName, testJson), testKey, testValue)
+			e2e.EnvNoError(s.t, env, err)
+		}
+		s.t.Log("Inserted 10 rows into the source table")
+
+		e2e.EnvWaitForEqualTables(env, s, "normalize inserts", dstTableName, "id,key,value,j")
+		env.CancelWorkflow()
+	}()
+
+	env.ExecuteWorkflow(peerflow.CDCFlowWorkflow, flowConnConfig, nil)
+	e2e.RequireEnvCanceled(s.t, env)
+}