The relative validity of adaptive and conventional testing strategies using non-test variables as one set of external criteria was investigated. A total of 101 college students completed both a variable length stradaptive test and peaked conventional test; a second group of 131 college students completed a variable length Bayesian adaptive test and the same peaked conventional test. All tests were computer-administered and consisted of five-alternative multiple-choice vocabulary items. Test scores were correlated with high school and college grade point averages and American College Testing Service Program subtest scores. The data showed generally higher criterion-related validities for the adaptive tests as compared to the conventional tests. In comparing the two adaptive testing procedures, the data suggested that the stradaptive test scored by mean difficulty methods resulted in more valid ability estimates than the Bayesian adaptive test. Conclusions must be considered tentative until supported by additional research. (RL)