VR filter not useable. If we went to +4 hours east, we could put it into position without problems. Test images on the sky at that point worked. However, upon slew to our targets at -1 hours west and +10 deg north declination, the VR filter slid out of position by itself. There was no warnings or errors triggered, but we could tell this was the case because the guider could not find a star and all stars on the image were severely out of focus donuts. There was no vignetting seen, so the filter completely moved out of the beam by itself. A switch to r at the same location worked and images were fine. This was repeatable, we tried a second time and the same thing happened, VR went in at +4hr east, but slid out upon slew to our science target without warnings or errors. We lost 1 hour trying to get the VR filter in and abandoned and used r for the rest of the night at about 2/3 normal throughput (approximately 1 hour lost) for a total of 2 lost hours.