This paper considers differences in modes of test administration, addressing three questions: (1) Do examinees respond to items in the same way across administration modes and computer interface variations? (2) What are some of the factors that can contribute to modal effects? and (3) Can item parameters calibrated from paper and pencil administrations be used for computer administrations? The questions were examined using data from paper and pencil and computer administrations of a fixed-form test in two different examinee samples. Several of the tests studied were complex. An initial comparability study was performed in 1998 involving approximately 8,600 students, and in response to that study, revisions were made to the computer interfaces. A follow-up study was performed in 2000 with approximately 12,000 examinees. This paper examines performance differences across paper and computer modes and across computer interface variations in both studies. Results are summarized at the total test level and for some individual items. Some factors that might have contributed to mode differences or affected computer performance in general are discussed. A small simulation study was also performed to examine the effect of using item parameters calibrated from paper and pencil administrations in a computer administration. Some items showed no performance differences across administration modes, but other items did. A variety of factors appeared to contribute to mode effects, and each item seemed to have a unique set of circumstances. Changes to the computer interface appeared to affect performance. Overall, results suggest that while performance effects do occur across modes, they have a fairly small effect in practice. Simulation results suggest that item parameters calibrated from paper and pencil tests could probably be used initially in a computer administration. (Contains 18 tables.) (SLD)