Auditory surveys are being used increasingly to monitor amphibians and assess amphibian declines. Interobserver differences in detected number of species during amphibian call surveys have not been widely studied, yet previous studies have assumed these differences are minimal. We examined inter-observer variation in 269 frog call surveys conducted along 20 standardized routes in central Texas. Two trained observers, listening simultaneously 10 m apart, agreed on number of species calling in 79.4% of 5-min surveys and 78.6% of 30-min surveys. The level of observer agreement varied among species. Observers in the present study were more likely to disagree about the presence of a species when only one or a few frogs called from distant sites. Wind and road noise had no significant effects on interobserver variation. Presence of moonlight was negatively associated with interobserver agreement. Because sampling variation and biases arising from observer effects may lead to inappropriate inferences and misdirected conservation efforts, it is important to control for interobserver differences during the design and analysis phases of research.