Abstract This study examines the magnitude of the effects of the computerized assessment format on MMPI responses, and discusses methodological considerations important in evaluating the sensitivity of personality inventories to different administration formats. Previous research suggested that subjects were more frank in impersonal assessment situations, and that computerized administration increases the use of the “Cannot Say” response. Although tending to minimize the effects of automated testing, the study designs and statistical procedures employed to date have been relatively insensitive to format differences. The present research analyzes item-level effects of computerized test administration using a crossover ANOVA design including the effects of order of administration, format of administration, and repeated testing. One hundred and fifty undergraduate subjects took a paper and pencil and computer-administered MMPI a week apart, in two different orders. Overall, the effects of the automated testing format are small compared with the effects of repeated testing when subjects; use of the “Cannot Say” response is controlled. Observed format differences are not in the direction of “frankness” or “candor” in the computer administration format. An index of each subject's sensitivity to computerized administration is computed and used to evaluate characteristics associated with shifting responses in the computerized testing format.