Our research objective focused on examining the instructional efficiency of tutoring as a form of instructional change as opposed to a non-tutoring approach in an outreach laboratory. We designed our laboratory based on cognitive load (CL) theory. Altogether, 269 twelfth-graders participated in our day-long module "Genetic Fingerprinting." In a quasi-experimental design, the control group (n = 121) followed the non-tutoring approach previously used, while the treatment group (n = 148) followed the newly developed tutoring approach. Each tutor was in charge of two student work groups and recorded the tutoring activities requested by the tutees throughout the day. We measured the students' invested mental effort (as an index of CL), cognitive achievement (in a pre-post-follow-up design), and the students' cooperation in their work groups as well as calculated the student instructional involvement (as a motivational variable). Additionally, we examined which aspects of the hands-on phases were of particular relevance to the students' invested mental effort. Unexpectedly, the combined mental effort and cognitive achievement data indicated that our implemented tutoring approach resulted in a lower instructional efficiency despite the relevance of tutoring for students' mental effort invested during the experimental phases. Most of the tutor assistance was unnecessarily requested for performing the procedural steps and using the equipment. Our results indicate an assistance dilemma and consequently underscore the necessity for effective tutor preparation in outreach laboratories.