Implementing laparoscopic surgery is a challenge that should prompt educational research in an attempt to establish a link between instruction and the quality of patient care. This randomized study was undertaken to compare the impact of instruction versus passive observation on laparoscopic skills. The task was stitching and tying a surgeon's knot on a perforated ulcer in a foam stomach placed in a simulator. Outcome measures were accuracy error, goal-directed and non-goal-directed actions, operating time, and tissue damage. Time/motion analysis was carried out by an masked assessor. Twelve participants were well matched for hand-eye coordination at pretesting. Regardless of whether instruction was given or not, a positive correlation was found between overall actions and operating time, non-goal-directed actions and operating time, and overall actions and non-goal directed actions. Intraoperative instruction decreased errors, but this study did not have sufficient power to detect small differences in other outcome measures. The teaching of minimal-access surgery should increasingly be based on educational research data rather than on unstructured rating by attending surgeons.