Three pyramidal adaptive tests and a conventional peaked test were constructed and administered by computer to two groups of students enrolled in undergraduate psychology courses. Six methods of scoring pyramidal tests were evaluated with respect to score distributions, stability, and the degree of relationship among scoring methods and between pyramidal scoring methods and scores on the conventional test. For both the pyramidal tests and the conventional test, score distributions were platkurtic and positively skewed. Two methods of scoring the pyramidal tests consistently used an equal or greater proportion of the range of possible scores then the conventional test. The 15-stage pyramidal tests showed test-retest correlations which were only slightly lower than that for the 40-item conventional test. However, when the effects of memory were considered, the pyramidal strategy yielded more stable ability estimates than conventional tests of equivalent length. The correlation between pyramidal test scores and those on conventional tests ranged from .82 to .86. One pair of scoring methods was found to be perfectly correlated for properly constructed pyramidal tests; a second pair correlated almost perfectly. (Author/SE)