Ecological monitoring programs typically aim to detect changes in the abundance of species of conservation concern or which reflect system status. Coral reef fish assemblages are functionally important for reef health and these are most commonly monitored using underwater visual surveys (UVS) by divers. In addition to estimating numbers, most programs also collect estimates of fish lengths to allow calculation of biomass, an important determinant of a fish’s functional impact. However, diver surveys may be biased because fishes may either avoid or are attracted to divers and the process of estimating fish length could result in fish counts that differ from those made without length estimations. Here we investigated whether (1) general diver disturbance and (2) the additional task of estimating fish lengths affected estimates of reef fish abundance and species richness during UVS, and for how long. Initial estimates of abundance and species richness were significantly higher than those made on the same section of reef after diver disturbance. However, there was no evidence that estimating fish lengths at the same time as abundance resulted in counts different from those made when estimating abundance alone. Similarly, there was little consistent bias among observers. Estimates of the time for fish taxa that avoided divers after initial contact to return to initial levels of abundance varied from three to 17 h, with one group of exploited fishes showing initial attraction to divers that declined over the study period. Our finding that many reef fishes may disperse for such long periods after initial contact with divers suggests that monitoring programs should take great care to minimise diver disturbance prior to surveys.