diff --git a/vllm/v1/attention/backends/mla/indexer.py b/vllm/v1/attention/backends/mla/indexer.py
index 7c0715a9e8b6..ed8c073e81e2 100644
--- a/vllm/v1/attention/backends/mla/indexer.py
+++ b/vllm/v1/attention/backends/mla/indexer.py
@@ -122,7 +122,10 @@ def get_name() -> str:
 
     @staticmethod
     def get_supported_kernel_block_sizes() -> list[int | MultipleOf]:
-        return [1, 64] if current_platform.is_rocm() else [64]
+        # Must not advertise 1 on ROCm: select_common_block_size picks
+        # min([1, 64]) = 1, which keeps block_size=1 and leaves the gluon
+        # preshuffle path (added in #41217) permanently disabled.
+        return [64]
 
     @classmethod
     def get_supported_head_sizes(cls) -> list[int]:
diff --git a/vllm/v1/attention/backends/mla/rocm_aiter_mla_sparse.py b/vllm/v1/attention/backends/mla/rocm_aiter_mla_sparse.py
index dc343b639f6c..095761704a15 100644
--- a/vllm/v1/attention/backends/mla/rocm_aiter_mla_sparse.py
+++ b/vllm/v1/attention/backends/mla/rocm_aiter_mla_sparse.py
@@ -273,7 +273,7 @@ class ROCMAiterMLASparseBackend(AttentionBackend):
 
     @staticmethod
     def get_supported_kernel_block_sizes() -> list[int | MultipleOf]:
-        return [1, 64]
+        return [64]
 
     @staticmethod
     def get_name() -> str: