Context: Trapping design influences information collected about wildlife populations and biodiversity. Trapping is also resource-intensive and has animal welfare implications. Aims: The scientific, financial and ethical performances of three trap designs were compared for estimating diversity and sampling small vertebrates. Methods: Small vertebrates were trapped over 16 trapping sessions, from April 2009 to May 2011, with aluminium box-style (Elliott) traps and two pitfall trap designs (shallow-wide and deep-narrow), in an arid environment. Key results: Shallow pitfalls recorded highest overall species richness (S = 22) and diversity (qD = 10.622), reptile diversity (qD = 8.112) and reptile capture rates (13.600 individuals per 100 trap nights). Shallow and deep pitfalls sampled ∼79.0% and 85.0% (respectively) more small mammals than did Elliott traps. Deep pitfalls sampled the greatest diversity (qD = 6.017) and number (29.700 individuals per 100 trap nights) of small mammals, and captured the greatest number of small mammal species (0.003) and individuals (0.106) per dollar. Shallow pitfalls were the most cost-efficient trap type for sampling reptile species (0.003) and individuals (0.044) per dollar. Between-session recapture rates were greatest in Elliott traps, indicating an increased likelihood of biased capture rates for certain small mammal species over time. Elliott traps were the least efficient traps on most scientific and cost measures, and recorded the greatest overall recapture rates, particularly for Sminthopsis crassicaudata and S. macroura. Body size of one species only, the nationally threatened Pseudomys australis, influenced its capture rate, with larger individuals more likely to be caught in deep pitfalls. Mortality was highest in pitfalls and mostly related to interactions between animals caught in the same trap. Key conclusions: Shallow pitfalls are suitable for studies focused on estimating species richness, and reptile diversity and abundance. Deep pitfalls are cost-effective for sampling small mammals. Ethical issues associated with pitfalls could be managed by checking traps more often at night, and/or including materials that provide increased protection from predators caught in the same trap, particularly during periods of high abundance. Implications: Trap design profoundly influences cost-effectiveness and welfare outcomes of wildlife research. We provide a tool to assist cost-benefit related decisions.