diff --git a/dlls/gdi32/dib.c b/dlls/gdi32/dib.c
index e8b3dc92200..d53b8f2e173 100644
--- a/dlls/gdi32/dib.c
+++ b/dlls/gdi32/dib.c
@@ -205,13 +205,12 @@ INT WINAPI StretchDIBits(HDC hdc, INT xDst, INT yDst, INT widthDst,
     }
     else /* use StretchBlt */
     {
-        HBITMAP hBitmap, hOldBitmap;
-        HPALETTE hpal = NULL;
-        HDC hdcMem;
         LONG height;
         LONG width;
         WORD planes, bpp;
         DWORD compr, size;
+        HBITMAP hBitmap;
+        BOOL fastpath = FALSE;
 
         release_dc_ptr( dc );
 
@@ -227,51 +226,84 @@ INT WINAPI StretchDIBits(HDC hdc, INT xDst, INT yDst, INT widthDst,
             return 0;
         }
 
-	hdcMem = CreateCompatibleDC( hdc );
-        hBitmap = CreateCompatibleBitmap(hdc, width, height);
-        hOldBitmap = SelectObject( hdcMem, hBitmap );
-        if(wUsage == DIB_PAL_COLORS)
+        hBitmap = GetCurrentObject(hdc, OBJ_BITMAP);
+
+        if (xDst == 0 && yDst == 0 && xSrc == 0 && ySrc == 0 &&
+            widthDst == widthSrc && heightDst == heightSrc &&
+            info->bmiHeader.biCompression == BI_RGB &&
+            dwRop == SRCCOPY)
         {
-            hpal = GetCurrentObject(hdc, OBJ_PAL);
-            hpal = SelectPalette(hdcMem, hpal, FALSE);
+            BITMAPOBJ *bmp;
+            if ((bmp = (BITMAPOBJ *)GDI_GetObjPtr( hBitmap, BITMAP_MAGIC )))
+            {
+                if (bmp->bitmap.bmBitsPixel == bpp &&
+                    bmp->bitmap.bmWidth == widthSrc &&
+                    bmp->bitmap.bmHeight == heightSrc &&
+                    bmp->bitmap.bmPlanes == planes)
+                    fastpath = TRUE;
+                GDI_ReleaseObj( hBitmap );
+            }
         }
 
-	if (info->bmiHeader.biCompression == BI_RLE4 ||
-	    info->bmiHeader.biCompression == BI_RLE8) {
-
-	   /* when RLE compression is used, there may be some gaps (ie the DIB doesn't
-	    * contain all the rectangle described in bmiHeader, but only part of it.
-	    * This mean that those undescribed pixels must be left untouched.
-	    * So, we first copy on a memory bitmap the current content of the
-	    * destination rectangle, blit the DIB bits on top of it - hence leaving
-	    * the gaps untouched -, and blitting the rectangle back.
-	    * This insure that gaps are untouched on the destination rectangle
-	    * Not doing so leads to trashed images (the gaps contain what was on the
-	    * memory bitmap => generally black or garbage)
-	    * Unfortunately, RLE DIBs without gaps will be slowed down. But this is
-	    * another speed vs correctness issue. Anyway, if speed is needed, then the
-	    * pStretchDIBits function shall be implemented.
-	    * ericP (2000/09/09)
-	    */
-
-            /* copy existing bitmap from destination dc */
-            StretchBlt( hdcMem, xSrc, abs(height) - heightSrc - ySrc,
-                        widthSrc, heightSrc, hdc, xDst, yDst, widthDst, heightDst,
-                        dwRop );
+        if (fastpath)
+        {
+            /* fast path */
+            TRACE("using fast path\n");
+            heightSrc = SetDIBits( hdc, hBitmap, 0, height, bits, info, wUsage);
         }
+        else
+        {
+            /* slow path - need to use StretchBlt */
+            HBITMAP hOldBitmap;
+            HPALETTE hpal = NULL;
+            HDC hdcMem;
 
-        SetDIBits(hdcMem, hBitmap, 0, height, bits, info, wUsage);
+            hdcMem = CreateCompatibleDC( hdc );
+            hBitmap = CreateCompatibleBitmap(hdc, width, height);
+            hOldBitmap = SelectObject( hdcMem, hBitmap );
+            if(wUsage == DIB_PAL_COLORS)
+            {
+                hpal = GetCurrentObject(hdc, OBJ_PAL);
+                hpal = SelectPalette(hdcMem, hpal, FALSE);
+            }
 
-        /* Origin for DIBitmap may be bottom left (positive biHeight) or top
-           left (negative biHeight) */
-        StretchBlt( hdc, xDst, yDst, widthDst, heightDst,
-		    hdcMem, xSrc, abs(height) - heightSrc - ySrc,
-		    widthSrc, heightSrc, dwRop );
-        if(hpal)
-            SelectPalette(hdcMem, hpal, FALSE);
-	SelectObject( hdcMem, hOldBitmap );
-	DeleteDC( hdcMem );
-	DeleteObject( hBitmap );
+            if (info->bmiHeader.biCompression == BI_RLE4 ||
+	            info->bmiHeader.biCompression == BI_RLE8) {
+
+                /* when RLE compression is used, there may be some gaps (ie the DIB doesn't
+                 * contain all the rectangle described in bmiHeader, but only part of it.
+                 * This mean that those undescribed pixels must be left untouched.
+                 * So, we first copy on a memory bitmap the current content of the
+                 * destination rectangle, blit the DIB bits on top of it - hence leaving
+                 * the gaps untouched -, and blitting the rectangle back.
+                 * This insure that gaps are untouched on the destination rectangle
+                 * Not doing so leads to trashed images (the gaps contain what was on the
+                 * memory bitmap => generally black or garbage)
+                 * Unfortunately, RLE DIBs without gaps will be slowed down. But this is
+                 * another speed vs correctness issue. Anyway, if speed is needed, then the
+                 * pStretchDIBits function shall be implemented.
+                 * ericP (2000/09/09)
+                 */
+
+                /* copy existing bitmap from destination dc */
+                StretchBlt( hdcMem, xSrc, abs(height) - heightSrc - ySrc,
+                            widthSrc, heightSrc, hdc, xDst, yDst, widthDst, heightDst,
+                            dwRop );
+            }
+
+            SetDIBits(hdcMem, hBitmap, 0, height, bits, info, wUsage);
+
+            /* Origin for DIBitmap may be bottom left (positive biHeight) or top
+               left (negative biHeight) */
+            StretchBlt( hdc, xDst, yDst, widthDst, heightDst,
+                hdcMem, xSrc, abs(height) - heightSrc - ySrc,
+                widthSrc, heightSrc, dwRop );
+            if(hpal)
+                SelectPalette(hdcMem, hpal, FALSE);
+            SelectObject( hdcMem, hOldBitmap );
+            DeleteDC( hdcMem );
+            DeleteObject( hBitmap );
+        }
     }
     return heightSrc;
 }