We trained four undergraduate students who reported no prior experience implementing behavior-analytic procedures to conduct a paired-stimulus preference assessment using a video self-instruction package. The package was composed of several components from prior research (i.e., a voice-over script, written instructions, multiple video models per step, and instructions for the trainee to rehearse throughout the video presentation). We used a nonconcurrent, multiple-baseline-across-participants design and found that all participants accurately implemented the preference assessment with a simulated client after viewing the video once. We discuss the contributions of the current results and directions for future research on video self-instruction for staff (i.e., maintenance, generalization, social validity).