Visual flyout count data for the common bent-wing bat Miniopterus schreibersii, collected by a team of observers over two seasons at a disused mine in the Kinglake National Park, south-eastern Australia, was compared with infra-red video footage, collected simultaneously, in order to quantify the precision and accuracy of the observer counts. Bayesian statistical models were used to evaluate the relationship between observer counts and the actual number of bats emerging from the cave, as determined by analysis of the infra-red video footage of the flyout. The accuracy of flyout counts was found to decline with increasing flyout rates, with observers' counts becoming increasingly negatively biased as the rate of bat emergence from the mine increased. In addition, there was evidence of inter-observer variation in the accuracy of the counts. Although the bias in observer counts was relatively small, caution needs to be exercised in interpreting the results of visual flyout counts. We conclude that the use of infra-red video footage for determining numbers is preferable to visual observer counts. The major difficulty in using flyout counts for monitoring is the considerable night-to-night variation in numbers of bats emerging, which could be attributed to variation in the proportion of bats emerging to forage, or to the use of alternative roosting sites by individual bats on successive nights. Both observer error and short-term temporal variation in numbers of emerging bats have the potential to bias population estimates of bats, and need to be properly accounted for if flyout counts are to be used as a tool for population assessment and monitoring.