A 15-stage pyramidal test and a 40-item two-stage test were constructed and administered by computer to 111 college undergraduates. The two-stage test was found to utilize a smaller proportion of its potential score range than the pyramidal test. Score distributions for both tests were positively skewed but not significantly different from the normal distribution. The pyramidal test's score distributions tended to be platykurtic while the two-stage test's distribution tended to be leptokurtic. The assignment of subjects to measurement subtests in the two-stage test was more accurate than in a previous empirical investigation since the misclassification rate was less than 1%. Comparison of scoring methods for the pyramidal strategy supported earlier findings that the average difficulty scoring methods were most useful. The correlations between scores on the two adaptive strategies ranged from r=.79 to .84. Both adaptive strategies appeared to adapt item difficulties to individual differences in abilities so as to reduce chance effects due to guessing. The pyramidal strategy seemed to be slightly more successful in eliminating guessing than the two-stage strategy. Results are discussed with respect to internal consistency reliabilities, stabilities, and the relation of each strategy to conventional testing. Simulation studies are suggested to further delineate the optimum characteristids of each testing strategy. (Author)