Background: Continuous neck and shoulder pain is a common musculoskeletal complaint. Physical exercise can reduce pain symptoms, but compliance to exercise is a challenge. Exercise-specific self-efficacy has been found to be a predictor of participation in preplanned exercise. Little is known about the influence of exercise-specific self-efficacy on compliance to workplace physical exercise. Purpose: To determine the influence of exercise-specific self-efficacy on compliance to specific strength exercises during working hours for laboratory technicians. Methods: We performed a cluster-randomized controlled trial, including laboratory technicians from two industrial production units in Copenhagen, Denmark. The participants were randomized to supervised specific strength exercises for the neck and shoulder muscles for 20 minutes three times a week (n = 282) or to a reference group (n = 255). The participants answered baseline and follow-up questions regarding self-efficacy and registered all exercises in a diary. Results: Overall compliance to exercises was 45 %. Compliance in company A (private sector) differed significantly between the three self-efficacy groups after 20 weeks. The odds ratio of compliance was 2.37 for moderate versus low self-efficacy, and 2.93 for high versus low self-efficacy. No significant difference was found in company B (public sector) or in the intervention group as a whole. Conclusion: We did not find self-efficacy to be a general statistically significant predictor of compliance to exercises during 20 weeks, but found self-efficacy to be a predictor of compliance in a private sector setting. Workplace-specific differences might be present and should be taken into account.