Recent articles on intervention research in occupational safety and health were reviewed. Representative publications, across a spectrum of industry types, were identified by computer search of the NIOSHTIC, Psychological Abstracts and Wilson Social Science Index databases. Articles summarized were classified as administrative, behavioral, or engineering, based on the type of intervention evaluated. Administrative interventions were defined as changes in work exposure or procedures, such as job rotation, rest schedules and work area monitoring. Behavioral interventions included safety training and other attempts to encourage safe work practices. Engineering interventions were physical changes in sources of occupational hazards or routes of exposure to hazards, such as improvements in ergonomics. Many studies lacked a theoretical foundation, relied on small samples or evaluated interventions lacking the power to produce desired outcomes. Studies that were not driven by theory offered no explanation of how an intervention was to produce an expected outcome, thus limiting understanding of how positive outcomes occurred and providing no conceptual framework for future improvements. The designs of most studies were quasiexperimental, having nonrandomly selected control and treatment groups or nonexperimental, in that they included only a treatment group and no controls. Although sample sizes varied among studies, none of the articles discussed recruitment of large enough samples to generate the statistical power needed to detect effects of interventions. The authors conclude that, since nonexperimental designs do not allow causal inference, quasiexperiments using nonrandom comparison groups or longitudinal data to study effects of intervention over time are better alternatives. Design limitations, such as quasiexperimentation and sample size issues should be discussed in publications. The authors recommend use of interdisciplinary research teams to conduct intervention research. Labor and management should be involved in planning the studies.