diff --git a/dlls/d3d9/cubetexture.c b/dlls/d3d9/cubetexture.c
index e140b2e103e..c3723143987 100644
--- a/dlls/d3d9/cubetexture.c
+++ b/dlls/d3d9/cubetexture.c
@@ -60,7 +60,7 @@ static ULONG WINAPI IDirect3DCubeTexture9Impl_Release(LPDIRECT3DCUBETEXTURE9 ifa
 
     if (ref == 0) {
         TRACE("Releasing child %p\n", This->wineD3DCubeTexture);
-        IWineD3DCubeTexture_Release(This->wineD3DCubeTexture);
+        IWineD3DCubeTexture_Destroy(This->wineD3DCubeTexture, D3D9CB_DestroySurface);
         IUnknown_Release(This->parentDevice);
         HeapFree(GetProcessHeap(), 0, This);
     }
diff --git a/dlls/d3d9/d3d9_private.h b/dlls/d3d9/d3d9_private.h
index 5aa7ea69505..0555b3a4b13 100644
--- a/dlls/d3d9/d3d9_private.h
+++ b/dlls/d3d9/d3d9_private.h
@@ -301,6 +301,9 @@ typedef struct IDirect3DSurface9Impl
 
     /* Parent reference */
     LPDIRECT3DDEVICE9       parentDevice;
+
+    /* If set forward refcounting to this object */
+    IUnknown                    *forwardReference;
 } IDirect3DSurface9Impl;
 
 /* ---------------------- */
@@ -543,4 +546,6 @@ extern HRESULT WINAPI D3D9CB_CreateRenderTarget(IUnknown *device, IUnknown *pSup
 
 extern ULONG WINAPI D3D9CB_DestroyDepthStencilSurface (IWineD3DSurface *pSurface);
 
+extern ULONG WINAPI D3D9CB_DestroySurface(IWineD3DSurface *pSurface);
+
 #endif /* __WINE_D3D9_PRIVATE_H */
diff --git a/dlls/d3d9/device.c b/dlls/d3d9/device.c
index bd96b10411c..82fe762266f 100644
--- a/dlls/d3d9/device.c
+++ b/dlls/d3d9/device.c
@@ -1011,8 +1011,20 @@ HRESULT WINAPI D3D9CB_CreateSurface(IUnknown *device, IUnknown *pSuperior, UINT
         *ppSurface = d3dSurface->wineD3DSurface;
         IUnknown_Release(d3dSurface->parentDevice);
         d3dSurface->parentDevice = NULL;
+        d3dSurface->forwardReference = pSuperior;
     } else {
         FIXME("(%p) IDirect3DDevice9_CreateSurface failed\n", device);
     }
     return res;
 }
+
+ULONG WINAPI D3D9CB_DestroySurface(IWineD3DSurface *pSurface) {
+    IDirect3DSurface9Impl* surfaceParent;
+    TRACE("(%p) call back\n", pSurface);
+
+    IWineD3DSurface_GetParent(pSurface, (IUnknown **) &surfaceParent);
+    /* GetParent's AddRef was forwarded to an object in destruction.
+     * Releasing it here again would cause an endless recursion. */
+    surfaceParent->forwardReference = NULL;
+    return IDirect3DSurface9_Release((IDirect3DSurface9*) surfaceParent);
+}
diff --git a/dlls/d3d9/surface.c b/dlls/d3d9/surface.c
index e95354049d9..c2f94a691c2 100644
--- a/dlls/d3d9/surface.c
+++ b/dlls/d3d9/surface.c
@@ -43,15 +43,13 @@ static HRESULT WINAPI IDirect3DSurface9Impl_QueryInterface(LPDIRECT3DSURFACE9 if
 
 static ULONG WINAPI IDirect3DSurface9Impl_AddRef(LPDIRECT3DSURFACE9 iface) {
     IDirect3DSurface9Impl *This = (IDirect3DSurface9Impl *)iface;
-    IUnknown *containerParent = NULL;
 
     TRACE("(%p)\n", This);
 
-    IWineD3DSurface_GetContainerParent(This->wineD3DSurface, &containerParent);
-    if (containerParent) {
-        /* Forward to the containerParent */
-        TRACE("(%p) : Forwarding to %p\n", This, containerParent);
-        return IUnknown_AddRef(containerParent);
+    if (This->forwardReference) {
+        /* Forward refcounting */
+        TRACE("(%p) : Forwarding to %p\n", This, This->forwardReference);
+        return IUnknown_AddRef(This->forwardReference);
     } else {
         /* No container, handle our own refcounting */
         ULONG ref = InterlockedIncrement(&This->ref);
@@ -64,15 +62,13 @@ static ULONG WINAPI IDirect3DSurface9Impl_AddRef(LPDIRECT3DSURFACE9 iface) {
 
 static ULONG WINAPI IDirect3DSurface9Impl_Release(LPDIRECT3DSURFACE9 iface) {
     IDirect3DSurface9Impl *This = (IDirect3DSurface9Impl *)iface;
-    IUnknown *containerParent = NULL;
 
     TRACE("(%p)\n", This);
 
-    IWineD3DSurface_GetContainerParent(This->wineD3DSurface, &containerParent);
-    if (containerParent) {
+    if (This->forwardReference) {
         /* Forward to the containerParent */
-        TRACE("(%p) : Forwarding to %p\n", This, containerParent);
-        return IUnknown_Release(containerParent);
+        TRACE("(%p) : Forwarding to %p\n", This, This->forwardReference);
+        return IUnknown_Release(This->forwardReference);
     } else {
         /* No container, handle our own refcounting */
         ULONG ref = InterlockedDecrement(&This->ref);
diff --git a/dlls/d3d9/tests/device.c b/dlls/d3d9/tests/device.c
index a1336f219c7..de8f66196c8 100644
--- a/dlls/d3d9/tests/device.c
+++ b/dlls/d3d9/tests/device.c
@@ -540,7 +540,7 @@ static void test_refcount(void)
         /* check implicit back buffer */
         hr = IDirect3DSwapChain9_GetBackBuffer(pSwapChain, 0, 0, &pBackBuffer);
         todo_wine CHECK_CALL( hr, "GetBackBuffer", pDevice, ++refcount);
-        todo_wine CHECK_REFCOUNT( pSwapChain, 1);
+        CHECK_REFCOUNT( pSwapChain, 1);
         if(pBackBuffer)
         {
             CHECK_SURFACE_CONTAINER( pBackBuffer, IID_IDirect3DSwapChain9, pSwapChain);
diff --git a/dlls/d3d9/texture.c b/dlls/d3d9/texture.c
index 427a7a2d80e..7eecacd3d79 100644
--- a/dlls/d3d9/texture.c
+++ b/dlls/d3d9/texture.c
@@ -59,7 +59,7 @@ static ULONG WINAPI IDirect3DTexture9Impl_Release(LPDIRECT3DTEXTURE9 iface) {
     TRACE("(%p) : ReleaseRef to %d\n", This, ref);
 
     if (ref == 0) {
-        IWineD3DTexture_Release(This->wineD3DTexture);
+        IWineD3DTexture_Destroy(This->wineD3DTexture, D3D9CB_DestroySurface);
         IUnknown_Release(This->parentDevice);
         HeapFree(GetProcessHeap(), 0, This);
     }