The Health Belief Model (HBM) relates a socio-psychologic theory of decision making to individual health-related behaviors. We conducted a meta-analysis of the relationships between four HBM dimensions (Susceptibility, Severity, Benefits and Costs) and health behavior on 16 studies that measured all four of the dimensions, measured a behavioral dependent variable and included some measures of reliability, minimal criteria for establishing the validity of the dimensions. Mean effect sizes were computed for all the studies, subgroupings representing studies of screening, risk reduction and adherence to medical regimen, and prospective and retrospective study designs. Of 24 mean effect sizes, 22 were found to be positive and statistically significant. The actual variance accounted for ranged from 0.001 to 0.09. Homogeneity was rejected for 15 of the 22, however, suggesting that the same underlying construct was not measured. Retrospective studies were found to have significantly large effect sizes for benefits and costs and smaller effect sizes for severity when compared to prospective studies. The weak effect sizes and lack of homogeneity indicate that it is premature to draw conclusions about the predictive validity of the HBM as operationalized in these studies. Our finding of only 16 studies meeting minimal criteria for valid representation of the HBM dimensions indicates that future studies should focus more on such issues.