diff --git a/mlir/include/mlir/Dialect/Tensor/Utils/Utils.h b/mlir/include/mlir/Dialect/Tensor/Utils/Utils.h
--- a/mlir/include/mlir/Dialect/Tensor/Utils/Utils.h
+++ b/mlir/include/mlir/Dialect/Tensor/Utils/Utils.h
@@ -38,7 +38,7 @@
 bool isCastLikeInsertSliceOp(InsertSliceOp op);
 
 /// A tensor.extract_slice is a cast-like operation if it merely rank-reduces
-/// the source tensor or extracts the entire source tensor.
+/// unit dimensions of the source tensor or extracts the entire source tensor.
 bool isCastLikeExtractSliceOp(ExtractSliceOp op);
 
 } // namespace tensor
diff --git a/mlir/lib/Dialect/Tensor/Utils/Utils.cpp b/mlir/lib/Dialect/Tensor/Utils/Utils.cpp
--- a/mlir/lib/Dialect/Tensor/Utils/Utils.cpp
+++ b/mlir/lib/Dialect/Tensor/Utils/Utils.cpp
@@ -98,8 +98,13 @@
   int64_t resultDim = 0;
   // Source dims and result dims (apart from dropped dims) must have the same
   // size.
-  for (int64_t dim = 0; dim < op.getSourceType().getRank(); ++dim) {
+  RankedTensorType sourceType = op.getSourceType();
+  for (int64_t dim = 0, e = sourceType.getRank(); dim < e; ++dim) {
     if (droppedDims.test(dim)) {
+      // ExtractSlice may drop unit dimensions that result from taking a size-1
+      // slice from a non-size-1 source dimension.
+      if (sourceType.getDimSize(dim) != 1)
+        return false;
       continue;
     }
     FailureOr<bool> equalDimSize = ValueBoundsConstraintSet::areEqual(
diff --git a/mlir/test/Dialect/Tensor/tracking-listener.mlir b/mlir/test/Dialect/Tensor/tracking-listener.mlir
--- a/mlir/test/Dialect/Tensor/tracking-listener.mlir
+++ b/mlir/test/Dialect/Tensor/tracking-listener.mlir
@@ -140,3 +140,14 @@
       {replacement_0 = 0} : tensor<1x5x1x1xf32> to tensor<3xf32>
   return
 }
+
+// -----
+
+func.func @non_cast_like_extract_slice_drop_non_unit_dim() {
+  // expected-error @below {{listener could not find replacement op}}
+  %0 = "test.foo"() {replaced} : () -> (tensor<f32>)
+  %1 = "test.foo"() : () -> (tensor<1x5x1x1xf32>)
+  %2 = tensor.extract_slice %1[0, 0, 0, 0][1, 1, 1, 1][1, 1, 1, 1]
+      {replacement_0 = 0} : tensor<1x5x1x1xf32> to tensor<f32>
+  return
+}