Underwater visual census (UVC) is a widely used technique for estimating species richness, abundance, and lengths of reef fishes. The technique has the advantage of being non-destructive, and can therefore be used for monitoring in marine protected areas. However, acquisition of robust data using UVC is reliant on the ability of individual divers to accurately identify fish, and estimate their lengths. Both of these variables can be affected by observer bias, which may also differ among observers. This study examines a technique with the potential to quantify and correct for observer bias in individual divers. We used simultaneous diver-operated stereo video and underwater visual census surveys, and compared the results to generate diver-specific correction factors. These correction factors were subsequently applied in additional estimates of fish length, with measurable improvement in the accuracy of the data.