Knowledge of the probability of detecting a species with a given survey effort is important when planning monitoring programs. The present paper compares detection probabilities for six survey methods: audio-visual transect surveys targeting adult frogs, and sweep netting, bait traps, small fyke nets, large fyke nets and seine netting targeting tadpoles. Data were drawn from repeat surveys of 16 wetlands over a 3-year period (2008-2011). Detection probabilities of each method and species combination were determined by maximum likelihood estimation. Audio-visual surveys targeting adults had a higher probability of detecting individual species than did methods targeting their tadpoles. However, the presence of adults was a reasonably poor indicator of the presence of their tadpoles on any given survey occasion. Of the methods employed to target tadpoles, large and small fyke nets had a significantly higher probability of detecting tadpoles of all species than did bait traps, seining or sweep nets. Differences in the detection probability of different methods demonstrated that the choice of survey method can bias towards particular groups of species and, in some cases, underestimate breeding activity by frog species in large waterbodies.