A number of strategies are described for adapting ability test items to individual differences in ability levels of testees. Each strategy consists of a different set of rules for selecting the sequence of test items to be administered to a given testee. Advantages and disadvantages of each strategy are discussed, and research issues unique to the strategy are described. Strategies reviewed are differentiated into two-stage approaches and multi-stage approaches. Several variations of the two-stage approach are described. Multi-stage strategies include fixed branching and variable branching strategies. Fixed branching strategies reviewed include a number of variations of the pyramidal approach (e.g., constant step size pyramids, decreasing step size pyramids, truncated pyramids, multiple-item pyramids), the flexilevel test, and the stradaptive test. Variable branching approaches include two Bayesian strategies and two maximum likelihood strategies. The various strategies are compared with each other on important characteristics and on practical considerations, and ranked on their apparent potential for providing measurement of equal precision at all levels of ability. (Author)