diff --git a/docs/envvars.rst b/docs/envvars.rst
index 466fae9f44..97eaed5ddc 100644
--- a/docs/envvars.rst
+++ b/docs/envvars.rst
@@ -379,6 +379,16 @@ Torch Compilation and Fusion
 LayerNorm/RMSNorm SM Margins
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
+.. envvar:: NVTE_CUDNN_MXFP8_NORM_OUTPUT_IN_INPUT_DTYPE
+
+   :Type: ``int`` (0 or 1)
+   :Default: ``0``
+   :Description: With cuDNN 9.25.0 or later, use the normalization input datatype for the virtual
+                 LayerNorm/RMSNorm output consumed by cuDNN MXFP8 block-scale quantization. This
+                 enables cuDNN's fused MXFP8 normalization engine, which requires matching FP16 or
+                 BF16 input and normalization-output datatypes. When set to ``0``, or with an
+                 earlier cuDNN version, the virtual normalization output uses FP32.
+
 .. envvar:: NVTE_FWD_LAYERNORM_SM_MARGIN
 
    :Type: ``int``
diff --git a/transformer_engine/common/normalization/common.cpp b/transformer_engine/common/normalization/common.cpp
index 375b109c23..3d00760b52 100644
--- a/transformer_engine/common/normalization/common.cpp
+++ b/transformer_engine/common/normalization/common.cpp
@@ -304,7 +304,17 @@ CudnnNormalizationPlan::CudnnNormalizationPlan(NVTE_Norm_Type NormType, NVTE_Nor
 
     if (_training) _rsigma->set_output(true).set_data_type(get_cudnn_fe_dtype(ctype));
 
-    const auto ZDtype = _fp8_out ? ctype : otype;
+    auto ZDtype = _fp8_out ? ctype : otype;
+    if (_fp8_out) {
+      const bool use_input_dtype = cudnnGetVersion() >= 92500 && _ndim_scale_block == 1 &&
+                                   use_cudnn_mxfp8_norm_output_in_input_dtype();
+      if (use_input_dtype) {
+        NVTE_WARN(
+            "The cuDNN MXFP8 normalization intermediate output uses the input dtype (itype) "
+            "instead of the compute dtype; otype still applies to the final quantized output.");
+        ZDtype = itype;
+      }
+    }
     _z->set_output(!_fp8_out).set_data_type(get_cudnn_fe_dtype(ZDtype));
 
     if (_fp8_out) {
@@ -562,6 +572,12 @@ bool& _zero_centered_gamma_in_weight_dtype() {
 
 bool& use_zero_centered_gamma_in_weight_dtype() { return _zero_centered_gamma_in_weight_dtype(); }
 
+bool use_cudnn_mxfp8_norm_output_in_input_dtype() {
+  static bool flag =
+      transformer_engine::getenv<bool>("NVTE_CUDNN_MXFP8_NORM_OUTPUT_IN_INPUT_DTYPE");
+  return flag;
+}
+
 }  //  namespace normalization
 }  // namespace transformer_engine
 
diff --git a/transformer_engine/common/normalization/common.h b/transformer_engine/common/normalization/common.h
index 31a547f62c..dd2f3d1459 100644
--- a/transformer_engine/common/normalization/common.h
+++ b/transformer_engine/common/normalization/common.h
@@ -308,6 +308,7 @@ bool use_cudnn_norm_fwd();
 bool use_cudnn_norm_bwd();
 
 bool& use_zero_centered_gamma_in_weight_dtype();
+bool use_cudnn_mxfp8_norm_output_in_input_dtype();
 
 }  // namespace normalization
 }  // namespace transformer_engine