diff --git a/table/position_delta_writer.go b/table/position_delta_writer.go
index e7c97541f..14dc00a2e 100644
--- a/table/position_delta_writer.go
+++ b/table/position_delta_writer.go
@@ -114,13 +114,16 @@ func (w *PositionDeltaWriter) Reinsert(batch arrow.RecordBatch) error {
 		return fmt.Errorf("%w: writer is already closed", ErrInvalidOperation)
 	}
 
-	indices := batch.Schema().FieldIndices(iceberg.RowIDColumnName)
-	if len(indices) == 0 {
+	rowIDIndex, found, err := rowIDColumnIndex(batch.Schema())
+	if err != nil {
+		return err
+	}
+	if !found {
 		return fmt.Errorf("%w: Reinsert batch must contain %s column",
 			iceberg.ErrInvalidArgument, iceberg.RowIDColumnName)
 	}
 
-	col := batch.Column(indices[0])
+	col := batch.Column(rowIDIndex)
 	if col.NullN() > 0 {
 		return fmt.Errorf("%w: Reinsert batch %s column must not contain null values",
 			iceberg.ErrInvalidArgument, iceberg.RowIDColumnName)
@@ -140,8 +143,12 @@ func (w *PositionDeltaWriter) Insert(batch arrow.RecordBatch) error {
 		return fmt.Errorf("%w: writer is already closed", ErrInvalidOperation)
 	}
 
-	if indices := batch.Schema().FieldIndices(iceberg.RowIDColumnName); len(indices) > 0 {
-		if batch.Column(indices[0]).NullN() != int(batch.NumRows()) {
+	rowIDIndex, found, err := rowIDColumnIndex(batch.Schema())
+	if err != nil {
+		return err
+	}
+	if found {
+		if batch.Column(rowIDIndex).NullN() != int(batch.NumRows()) {
 			return fmt.Errorf("%w: Insert batch %s column must be all null (use Reinsert for preserved IDs)",
 				iceberg.ErrInvalidArgument, iceberg.RowIDColumnName)
 		}
@@ -245,7 +252,11 @@ func (w *PositionDeltaWriter) buildUnifiedIterator() iter.Seq2[arrow.RecordBatch
 // already has _row_id (validated to be all-null by Insert), it is returned
 // retained as-is.
 func appendNullRowIDColumn(alloc memory.Allocator, batch arrow.RecordBatch) (arrow.RecordBatch, error) {
-	if indices := batch.Schema().FieldIndices(iceberg.RowIDColumnName); len(indices) > 0 {
+	_, found, err := rowIDColumnIndex(batch.Schema())
+	if err != nil {
+		return nil, err
+	}
+	if found {
 		batch.Retain()
 
 		return batch, nil
@@ -278,3 +289,16 @@ func appendNullRowIDColumn(alloc memory.Allocator, batch arrow.RecordBatch) (arr
 
 	return array.NewRecordBatch(newSchema, cols, nrows), nil
 }
+
+func rowIDColumnIndex(schema *arrow.Schema) (int, bool, error) {
+	indices := schema.FieldIndices(iceberg.RowIDColumnName)
+	switch len(indices) {
+	case 0:
+		return 0, false, nil
+	case 1:
+		return indices[0], true, nil
+	default:
+		return 0, false, fmt.Errorf("%w: batch contains multiple %s columns",
+			iceberg.ErrInvalidArgument, iceberg.RowIDColumnName)
+	}
+}
diff --git a/table/position_delta_writer_test.go b/table/position_delta_writer_test.go
index 55eb524a0..629d3931a 100644
--- a/table/position_delta_writer_test.go
+++ b/table/position_delta_writer_test.go
@@ -280,3 +280,49 @@ func TestPositionDeltaWriter_ReinsertRejectsMissingColumn(t *testing.T) {
 	require.Error(t, err)
 	assert.Contains(t, err.Error(), "must contain _row_id")
 }
+
+func TestPositionDeltaWriter_RejectsDuplicateRowIDColumns(t *testing.T) {
+	for _, layout := range []struct {
+		name       string
+		appendRows func(first, second *array.Int64Builder)
+	}{
+		{"both null", func(first, second *array.Int64Builder) {
+			first.AppendNull()
+			second.AppendNull()
+		}},
+		{"both non-null", func(first, second *array.Int64Builder) {
+			first.Append(10)
+			second.Append(11)
+		}},
+	} {
+		t.Run(layout.name, func(t *testing.T) {
+			schema := arrow.NewSchema([]arrow.Field{
+				{Name: "id", Type: arrow.PrimitiveTypes.Int64},
+				{Name: iceberg.RowIDColumnName, Type: arrow.PrimitiveTypes.Int64, Nullable: true},
+				{Name: iceberg.RowIDColumnName, Type: arrow.PrimitiveTypes.Int64, Nullable: true},
+			}, nil)
+			builder := array.NewRecordBuilder(memory.DefaultAllocator, schema)
+			defer builder.Release()
+			builder.Field(0).(*array.Int64Builder).Append(1)
+			layout.appendRows(builder.Field(1).(*array.Int64Builder), builder.Field(2).(*array.Int64Builder))
+			batch := builder.NewRecordBatch()
+			defer batch.Release()
+
+			for _, operation := range []struct {
+				name string
+				call func(*table.PositionDeltaWriter) error
+			}{
+				{"Reinsert", func(w *table.PositionDeltaWriter) error { return w.Reinsert(batch) }},
+				{"Insert", func(w *table.PositionDeltaWriter) error { return w.Insert(batch) }},
+			} {
+				t.Run(operation.name, func(t *testing.T) {
+					writer, err := table.NewPositionDeltaWriter(newV3RowLineageTestTable(t))
+					require.NoError(t, err)
+					err = operation.call(writer)
+					require.ErrorIs(t, err, iceberg.ErrInvalidArgument)
+					require.ErrorContains(t, err, "multiple _row_id columns")
+				})
+			}
+		})
+	}
+}