The limitations people with profound intellectual disabilities experience in functioning contribute to a vulnerability to self-injurious behavior. Since this problem behavior has important negative consequences for people concerned, examining the effectiveness of treatments is important. In the current meta-analysis, single-case studies investigating non-aversive, non-intrusive forms of reinforcement were combined using hierarchical linear models. This analysis revealed that the average effect of treatment was relatively large and statistically significant. Further, significant variance was observed between both studies and participants. In addition, some evidence was found for a moderator effect of sensory impairment. Finally, no statistically significant moderator effects of medication, motor impairment, setting, age, gender, matching of treatment with behavioral function and contingency were found.