Artificial refuges (cover boards) are commonly used to survey and monitor herpetofauna in many parts of the world. Despite the extensive use of artificial refuges in mesic environments, their effectiveness for detecting amphibians in temperate zones has rarely been examined. We compared amphibian detection probabilities between two survey methods; active searches of natural habitat and artificial refuges of three different types (corrugated steel, roofing tiles and timber railway sleepers). Our study area included five bioregions encompassing a 1180-km latitudinal gradient across a modified, temperate eucalypt woodland vegetation community in south-eastern Australia. We deployed 14 778 artificial refuges in terrestrial environments, within patches of remnant vegetation, and collected presence and abundance data on herpetofauna between 1999 and 2017. We used Bayesian logistic regression to identify the most effective survey method for detecting frog species across all bioregions. We modelled frog detections by fitting survey method, time since refuge deployment and rainfall prior to each survey. We detected 3970 individuals from 18 frog species. Overall, we found active searches and timber substrates most effective for detecting a broad range of species, although detection rates were driven by the numerically abundant spotted marsh frog Limnodynastes tasmaniensis. Timber refuges were effective for detecting several burrowing species, whereas active searches were effective at detecting habitat generalists. Quadratic effects of rainfall prior to survey as opposed to linear effects of time since artificial refuge placement was important in explaining frog detection rates in some bioregions. Active searches, timber railway sleepers and sheets of corrugated steel provide complimentary survey methods for detecting amphibians, although detection rates are influenced by rainfall patterns. Artificial refuges provide a time-effective and standardized method for studying amphibians in their non-breeding terrestrial environment and should be incorporated into future surveys and biodiversity monitoring programmes.