With recent increases in the numbers of species reintroduction projects and reintroduction-related publications, there is now a recognizable field of reintroduction biology. Nevertheless, research thus far has been fragmented and ad hoc, rather than an organized attempt to gain reliable knowledge to improve reintroduction success. We reviewed 454 recent (1990-2005) peer-reviewed papers dealing with wildlife reintroductions from 101 journals. Most research has been retrospective, either opportunistic evaluations of techniques or general project summaries, and most inference is gained from post hoc interpretation of monitoring results on a species-by-species basis. Documentation of reintroduction outcomes has improved, however, and the derivation of more general principles via meta-analyses is expected to increase. The fragmentation of the reintroduction literature remains an obstacle. There is scope to improve reintroduction biology by greater application of the hypothetico-deductive method, particularly through the use of modeling approaches and well-designed experiments. Examples of fruitful approaches in reintroduction research include experimental studies to improve outcomes from the release of captive-bred animals, use of simulation modeling to identify factors affecting the viability of reintroduced populations, and the application of spatially explicit models to plan for and evaluate reintroductions. We recommend that researchers contemplating future reintroductions carefully determine a priori the specific goals, overall ecological purpose, and inherent technical and biological limitations of a given reintroduction and that evaluation processes incorporate both experimental and modeling approaches. We suggest that the best progress will be made when multidisciplinary teams of resource managers and scientists work in close collaboration and when results from comparative analyses, experiments, and modeling are combined within and among studies.