Population monitoring of seabirds plays an important role in conservation since it provides the information required to evaluate conservation programmes of endangered species, to guide harvest management and to monitor indicators of marine ecosystem health. Annual nest counts are often used for the long-term monitoring of breeding seabird populations. While such counts provide a direct and cost-effective survey method, single nest counts will almost always yield an underestimate of the true number of nests and provide no means of expressing uncertainty. In this study, we used double counts and the Lincoln-Petersen estimator to assess the reliability of nest counts for the Yellow-Eyed Penguin (Megadyptes antipodes) in the South Island of New Zealand. We estimated the detection rate of single nest counts at 88%, and we calculated that double counts are sufficiently precise to detect changes in the size of the breeding population as small as 3.3% between years. We failed to detect observer bias or any effects of search effort or habitat on the detection rate, the only significant factor being the steepness of the terrain. Given the high detection rate and the susceptibility of Yellow-Eyed Penguins to human disturbance, we conclude that the established protocol for single nest counts is adequate for the regular monitoring of the population. We suggest that double counts could be conducted once every 5 years to obtain precise estimates for the purpose of longterm population monitoring. We also highlight the importance of repeating similar studies for a variety of species and habitats.