Insectivorous bats were surveyed using harp traps and ultrasonic bat detectors at 284 sites in three regions of Victoria, south-eastern Australia. Ten species were recorded in the Box-Ironbark region, 16 species in North-east Victoria, and 14 species in Gippsland. Detectors yielded more species records per site in the Box-Ironbark region (where forests are dry, open and fragmented), whereas harp traps were more successful in the North-east and Gippsland (where forests are generally denser and taller). Results suggest that detectors are likely to be most effective where there are fewer potential species to be considered in call identification, and that harp traps are likely to be most effective in areas of relatively dense vegetation with discrete flyways. Overall the two techniques were complementary, as they were subject to different biases in the suite of species most likely to be recorded. A library of over 250 ultrasonic reference calls was compiled as the basis for the identification of unknown calls. There was a high level of variation in the search-phase calls produced by each species, and between 15 and 40 reference calls were considered necessary to gain an adequate representation of this variation within the study area. This high level of intra-specific variability, combined with a considerable inter-specific overlap in call characteristics, resulted in a conservative approach to call identification. The development of an identification key was useful in alleviating problems of subjectivity and lack of repeatability associated with qualitative methods of call identification. Investigation of optimal sampling periods revealed that between three and five hours of recording after sunset was required to obtain records of approximately 70% of the species identified from a complete night of detector data. Suggestions are provided for the design of future surveys using harp traps and detectors.