Estimating abundance is important in many ecological studies in order to understand the spatial and temporal dynamics of a population, which can assist in management and conservation. However, direct estimates of abundance can be difficult and expensive to obtain, particularly for wide-ranging, rare or elusive species. An alternative Á estimating from detection-nondetection data Á is a challenging but alluring concept to ecologists since the cost and effort of a study can be greatly reduced. This paper describes a method for estimating the abundance of randomly distributed or aggregated populations by using binary data where the probability of detection is less than one. The performances of the models were evaluated by computer simulations comprising 1620 cases. The results show that the accuracy of the abundance estimates increases as the sampling rate, efficiency of survey method, and the number of repeated surveys increase, whereas the accuracy declines as individuals become more aggregated. For a randomly distributed population, using a sampling rate of 0.05 in a survey method with a detection probability of 0.5, and repeating surveys three times provides sufficient accuracy of abundance. For an aggregated population, to achieve reasonably accurate abundance estimates the sampling rate should be doubled and each cell should be repeatedly surveyed on 4 to 6 occasions.