diff --git a/include/cglm/mat4.h b/include/cglm/mat4.h
index cda5285..c099574 100644
--- a/include/cglm/mat4.h
+++ b/include/cglm/mat4.h
@@ -539,7 +539,9 @@ glm_mat4_scale_p(mat4 m, float s) {
 CGLM_INLINE
 void
 glm_mat4_scale(mat4 m, float s) {
-#if defined( __SSE__ ) || defined( __SSE2__ )
+#ifdef __AVX__
+  glm_mat4_scale_avx(m, s);
+#elif defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_scale_sse2(m, s);
 #elif defined(CGLM_NEON_FP)
   glm_mat4_scale_neon(m, s);
diff --git a/include/cglm/simd/avx/mat4.h b/include/cglm/simd/avx/mat4.h
index 944769b..e8c36c8 100644
--- a/include/cglm/simd/avx/mat4.h
+++ b/include/cglm/simd/avx/mat4.h
@@ -14,6 +14,16 @@
 
 #include <immintrin.h>
 
+CGLM_INLINE
+void
+glm_mat4_scale_avx(mat4 m, float s) {
+  __m256 y0;
+  y0 = _mm256_set1_ps(s);
+  
+  glmm_store256(m[0], _mm256_mul_ps(y0, glmm_load256(m[0])));
+  glmm_store256(m[2], _mm256_mul_ps(y0, glmm_load256(m[2])));
+}
+
 CGLM_INLINE
 void
 glm_mat4_mul_avx(mat4 m1, mat4 m2, mat4 dest) {