diff --git a/table/arrow_scanner.go b/table/arrow_scanner.go
index 26b18d191..02da7fa1d 100644
--- a/table/arrow_scanner.go
+++ b/table/arrow_scanner.go
@@ -391,6 +391,32 @@ func groupPosDeletesByFilePath(ctx context.Context, filePathCol, posCol *arrow.C
 	return results, nil
 }
 
+func collectPosDeletePositions(positionalDeletes positionDeletes) (set[int64], error) {
+	deletes := set[int64]{}
+	for _, chunk := range positionalDeletes {
+		if chunk == nil {
+			return nil, fmt.Errorf("%w: nil pos column chunk in position delete file",
+				iceberg.ErrInvalidSchema)
+		}
+		if chunk.DataType().ID() != arrow.INT64 {
+			return nil, fmt.Errorf("%w: unsupported pos column type %s in position delete file",
+				iceberg.ErrInvalidSchema, chunk.DataType())
+		}
+		for _, arr := range chunk.Chunks() {
+			posArr, ok := arr.(*array.Int64)
+			if !ok {
+				return nil, fmt.Errorf("%w: unsupported pos chunk array type %T in position delete file",
+					iceberg.ErrInvalidSchema, arr)
+			}
+			for _, v := range posArr.Int64Values() {
+				deletes[v] = struct{}{}
+			}
+		}
+	}
+
+	return deletes, nil
+}
+
 func releasePosDeletes(deletes map[string]*arrow.Chunked) {
 	for _, chunk := range deletes {
 		if chunk != nil {
@@ -1060,13 +1086,9 @@ func (as *arrowScan) recordsFromTask(ctx context.Context, task internal.Enumerat
 	}
 
 	if applyPosDeletes {
-		deletes := set[int64]{}
-		for _, chunk := range positionalDeletes {
-			for _, a := range chunk.Chunks() {
-				for _, v := range a.(*array.Int64).Int64Values() {
-					deletes[v] = struct{}{}
-				}
-			}
+		deletes, err := collectPosDeletePositions(positionalDeletes)
+		if err != nil {
+			return err
 		}
 
 		pipeline = append(pipeline, processPositionalDeletes(ctx, deletes, posSource.cursor()))
@@ -1151,13 +1173,9 @@ func (as *arrowScan) producePosDeletesFromTask(ctx context.Context, task interna
 	pipeline := make([]recProcessFn, 0, 2)
 	pipeline = append(pipeline, enrichRecordsWithPosDeleteFields(ctx, task.Value.File, posSource.cursor()))
 	if len(positionalDeletes) > 0 {
-		deletes := set[int64]{}
-		for _, chunk := range positionalDeletes {
-			for _, a := range chunk.Chunks() {
-				for _, v := range a.(*array.Int64).Int64Values() {
-					deletes[v] = struct{}{}
-				}
-			}
+		deletes, err := collectPosDeletePositions(positionalDeletes)
+		if err != nil {
+			return err
 		}
 
 		pipeline = append(pipeline, processPositionalDeletes(ctx, deletes, posSource.cursor()))
diff --git a/table/arrow_scanner_posdelete_regression_test.go b/table/arrow_scanner_posdelete_regression_test.go
index e9bca3e9f..955809bca 100644
--- a/table/arrow_scanner_posdelete_regression_test.go
+++ b/table/arrow_scanner_posdelete_regression_test.go
@@ -250,6 +250,21 @@ func TestGroupPosDeletesByFilePathRejectsUnsupportedFilePathLayout(t *testing.T)
 	}
 }
 
+func TestCollectPosDeletePositionsRejectsUnsupportedPosType(t *testing.T) {
+	mem := memory.NewCheckedAllocator(memory.DefaultAllocator)
+	defer mem.AssertSize(t, 0)
+
+	posArr := int32Array(mem, 1, 2)
+	defer posArr.Release()
+
+	posCol := arrow.NewChunked(arrow.PrimitiveTypes.Int32, []arrow.Array{posArr})
+	defer posCol.Release()
+
+	_, err := collectPosDeletePositions(positionDeletes{posCol})
+	require.ErrorIs(t, err, iceberg.ErrInvalidSchema)
+	assert.Contains(t, err.Error(), "unsupported pos column type")
+}
+
 // TestProcessPositionalDeletesAcrossBatches is the regression net for the
 // positional-delete index bug: processPositionalDeletes applies deletes one Arrow
 // batch at a time, but the surviving-row indices must index into the *current