Video motion analysis software was used by introductory physics students in a variety of instructional settings. 368 high school and college students took part in a study where the effect of graduated variations in the use of a video analysis tool was examined. Post-instruction assessment of student ability to interpret kinematics graphs indicates that groups using the tool generally performed better than students taught via traditional instruction. The data further establishes that the greater the integration of video analysis into the kinematics curriculum, the larger the educational impact. An additional comparison showed that graph interpretation skills were significantly better when a few traditional labs were simply replaced with video analysis experiments. Hands-on involvement appeared to play a critical role. Limiting student experience with the video analysis technique to a single teacher-led demonstration resulted in no improvement in performance relative to traditional instruction. Offering more extensive demonstrations and carrying them out over an extended period of time proved somewhat effective. The greatest impact came from a combination of demonstrations with hands-on labs. The curricular modifications employed in the different classrooms and the methods used to evaluate them are discussed.