PURPOSE This paper is a meta-analysis of data from in vitro studies and short-term animal studies that have combined extremely low frequency magnetic fields with known carcinogens or other toxic physical or chemical agents. MATERIALS AND METHODS The data was analyzed by systematic comparison of study characteristics between positive and negative studies to reveal possible consistent patterns. RESULTS The majority of the studies reviewed were positive, suggesting that magnetic fields do interact with other chemical and physical exposures. Publication bias is unlikely to explain the findings. Interestingly, a nonlinear 'dose-response' was found, showing a minimum percentage of positive studies at fields between 1 and 3 mT. The radical pair mechanism (magnetic field effects on recombination of radical pairs) is a good candidate mechanism for explaining the biphasic dose-response seen in the present analysis. CONCLUSIONS Most of the studies reviewed used magnetic fields of 100 microT or higher, so the findings are not directly relevant for explaining the epidemiological findings suggesting increased risk of childhood leukemia above 0.4 microT. However, confirmed adverse effects even at 100 microT would have implications for risk assessment and management, including the need to reconsider the exposure limits for magnetic fields. There is an obvious need for further studies on combined effects with magnetic fields.