diff --git a/GL/draw.c b/GL/draw.c
index 807d43a..b1f98ef 100644
--- a/GL/draw.c
+++ b/GL/draw.c
@@ -86,7 +86,9 @@ static GLboolean _glIsVertexDataFastPathCompatible() {
     if(UV_POINTER.size != 2) return GL_FALSE;
 
     if(DIFFUSE_POINTER.type != GL_UNSIGNED_BYTE) return GL_FALSE;
-    if(DIFFUSE_POINTER.size != 4) return GL_FALSE;
+
+    /* BGRA is the required color order */
+    if(DIFFUSE_POINTER.size != GL_BGRA) return GL_FALSE;
 
     return GL_TRUE;
 }
@@ -381,6 +383,30 @@ static void _readVertexData3ubARGB(const GLubyte* input, GLuint count, GLubyte s
     }
 }
 
+static void _readVertexData4ubRevARGB(const GLubyte* input, GLuint count, GLubyte stride, GLubyte* output) {
+    ITERATE(count) {
+        output[0] = input[0];
+        output[1] = input[1];
+        output[2] = input[2];
+        output[3] = input[3];
+
+        input += stride;
+        output += sizeof(Vertex);
+    }
+}
+
+static void _readVertexData4fRevARGB(const float* input, GLuint count, GLubyte stride, GLubyte* output) {
+    ITERATE(count) {
+        output[0] = (GLubyte) clamp(input[0] * 255.0f, 0, 255);
+        output[1] = (GLubyte) clamp(input[1] * 255.0f, 0, 255);
+        output[2] = (GLubyte) clamp(input[2] * 255.0f, 0, 255);
+        output[3] = (GLubyte) clamp(input[3] * 255.0f, 0, 255);
+
+        input = (float*) (((GLubyte*) input) + stride);
+        output += sizeof(Vertex);
+    }
+}
+
 static void _fillWithNegZVE(GLuint count, GLfloat* output) {
     ITERATE(count) {
         output[0] = output[1] = 0.0f;
@@ -430,6 +456,14 @@ static void _readVertexData4uiARGB(const GLuint* input, GLuint count, GLubyte st
     assert(0 && "Not Implemented");
 }
 
+static void _readVertexData4usRevARGB(const GLushort* input, GLuint count, GLubyte stride, GLubyte* output) {
+    assert(0 && "Not Implemented");
+}
+
+static void _readVertexData4uiRevARGB(const GLuint* input, GLuint count, GLubyte stride, GLubyte* output) {
+    assert(0 && "Not Implemented");
+}
+
 GLuint* _glGetEnabledAttributes() {
     return &ENABLED_VERTEX_ATTRIBUTES;
 }
@@ -792,7 +826,28 @@ static inline void _readDiffuseData(const GLuint first, const GLuint count, Vert
         default:
             assert(0 && "Not Implemented");
         }
-    } else {
+    } else if(DIFFUSE_POINTER.size == GL_BGRA) {
+        switch(DIFFUSE_POINTER.type) {
+            case GL_DOUBLE:
+            case GL_FLOAT:
+                _readVertexData4fRevARGB(cptr, count, cstride, output[0].bgra);
+            break;
+            case GL_BYTE:
+            case GL_UNSIGNED_BYTE:
+                _readVertexData4ubRevARGB(cptr, count, cstride, output[0].bgra);
+            break;
+            case GL_SHORT:
+            case GL_UNSIGNED_SHORT:
+                _readVertexData4usRevARGB(cptr, count, cstride, output[0].bgra);
+            break;
+            case GL_INT:
+            case GL_UNSIGNED_INT:
+                _readVertexData4uiRevARGB(cptr, count, cstride, output[0].bgra);
+            break;
+        default:
+            assert(0 && "Not Implemented");
+        }
+    }else {
         assert(0 && "Not Implemented");
     }
 }
@@ -802,6 +857,7 @@ static void generate(SubmissionTarget* target, const GLenum mode, const GLsizei
     /* Read from the client buffers and generate an array of ClipVertices */
     TRACE();
 
+    static const uint32_t FAST_PATH_BYTE_SIZE = (sizeof(GLfloat) * 3) + (sizeof(GLfloat) * 2) + (sizeof(GLubyte) * 4);
     const GLsizei istride = byte_size(type);
 
     if(!indices) {
@@ -815,12 +871,9 @@ static void generate(SubmissionTarget* target, const GLenum mode, const GLsizei
             Vertex* it = start;
             ITERATE(count) {
                 it->flags = PVR_CMD_VERTEX;
-
-                memcpy(it->xyz, pos, sizeof(GLfloat) * 3);
-                memcpy(it->uv, pos + 3, sizeof(GLfloat) * 2);
-                memcpy(it->bgra, pos + 5, sizeof(GLubyte) * 4);
-                pos += 32 / sizeof(GLfloat);
+                memcpy(it->xyz, pos, FAST_PATH_BYTE_SIZE);
                 it++;
+                pos += 32 / sizeof(GLfloat);
             }
         } else {
             _readPositionData(first, count, start);
@@ -1385,7 +1438,7 @@ void APIENTRY glVertexPointer(GLint size,  GLenum type,  GLsizei stride,  const
 void APIENTRY glColorPointer(GLint size,  GLenum type,  GLsizei stride,  const GLvoid * pointer) {
     TRACE();
 
-    if(size != 3 && size != 4) {
+    if(size != 3 && size != 4 && size != GL_BGRA) {
         _glKosThrowError(GL_INVALID_VALUE, __func__);
         _glKosPrintError();
         return;
diff --git a/GL/immediate.c b/GL/immediate.c
index eceb8e2..ae3ad7a 100644
--- a/GL/immediate.c
+++ b/GL/immediate.c
@@ -55,7 +55,7 @@ void _glInitImmediateMode(GLuint initial_size) {
     UV_ATTRIB.size = 2;
 
     DIFFUSE_ATTRIB.ptr = VERTEX_ATTRIB.ptr + (sizeof(GLfloat) * 5);
-    DIFFUSE_ATTRIB.size = 4;
+    DIFFUSE_ATTRIB.size = GL_BGRA;  /* Flipped color order */
     DIFFUSE_ATTRIB.type = GL_UNSIGNED_BYTE;
     DIFFUSE_ATTRIB.stride = 32;
 
@@ -152,10 +152,10 @@ void APIENTRY glVertex3f(GLfloat x, GLfloat y, GLfloat z) {
     vert->u = UV_COORD[0];
     vert->v = UV_COORD[1];
 
-    vert->rgba[0] = COLOR[0];
-    vert->rgba[1] = COLOR[1];
-    vert->rgba[2] = COLOR[2];
-    vert->rgba[3] = COLOR[3];
+    vert->bgra[R8IDX] = COLOR[0];
+    vert->bgra[G8IDX] = COLOR[1];
+    vert->bgra[B8IDX] = COLOR[2];
+    vert->bgra[A8IDX] = COLOR[3];
 
     memcpy(st, ST_COORD, sizeof(GLfloat) * 2);
     memcpy(n, NORMAL, sizeof(GLfloat) * 3);
diff --git a/GL/state.c b/GL/state.c
index ef026a5..239f5d9 100644
--- a/GL/state.c
+++ b/GL/state.c
@@ -671,7 +671,7 @@ const GLubyte *glGetString(GLenum name) {
             return (const GLubyte*) "1.2 (partial) - GLdc 1.1";
 
         case GL_EXTENSIONS:
-            return (const GLubyte*) "GL_ARB_framebuffer_object, GL_ARB_multitexture, GL_ARB_texture_rg, GL_EXT_paletted_texture, GL_EXT_shared_texture_palette, GL_KOS_multiple_shared_palette";
+            return (const GLubyte*) "GL_ARB_framebuffer_object, GL_ARB_multitexture, GL_ARB_texture_rg, GL_EXT_paletted_texture, GL_EXT_shared_texture_palette, GL_KOS_multiple_shared_palette, GL_ARB_vertex_array_bgra";
     }
 
     return (const GLubyte*) "GL_KOS_ERROR: ENUM Unsupported\n";
diff --git a/include/glkos.h b/include/glkos.h
index 082895c..5af510b 100644
--- a/include/glkos.h
+++ b/include/glkos.h
@@ -60,7 +60,7 @@ typedef struct {
     GLfloat z;
     GLfloat u;
     GLfloat v;
-    GLubyte rgba[4];
+    GLubyte bgra[4];
     GLuint padding1;
 } GLVertexKOS;