Research to date on adaptive (sequential, branched, individualized, tailored, programmed, response-contingent) ability testing is reviewed and summarized, following a brief review of problems inherent in conventional individual and group approaches to ability measurement. Research reviewed includes empirical, simulation and theoretical studies of adaptive testing strategies. Adaptive strategies identified in the literature include two-stage testing and multistage tests. Multistage tests are differentiated into fixed branching models and variable branching models (including Bayesian and non-Bayesian strategies). Results of research using the various strategies and research approaches are compared and summarized, leading to the general conclusion that adaptive testing can considerably reduce testing time and at the same time yield scores of higher reliability and validity than conventional tests, under a number of circumstances. A number of new psychometric problems raised by adaptive testing are discussed, as is the criterion problem in evaluating the utility of adaptive testing. Problems of implementing adaptive testing in a paper and pencil mode, or using special testing machines are reviewed; the advantages of computer-controlled adaptive test administration are described. (Author)