A male-female stacked version of Pender's Health Promotion Model (HPM) was used to evaluate whether the causal mechanisms underlying health promotion behaviors are the same for men and women. Indicators of the conceptual variables in the HPM were obtained from the National Survey of Personal Health Practices and Consequences, 1979-1980--a telephone sample survey of 3,025 noninstitutionalized adults. All coefficients to be estimated were initially constrained to be equal for men and women, except for six direct effects originating from marital status. The estimates obtained via the LISREL VI program indicated that the stacked model did not fit the data and that additional gender interactions were required between gender and other concepts within the HPM. Based on model diagnostics and with the relaxation of some equality constraints, a modified stacked model was constructed that fit the data. The explanatory power of the model was poor, calling into question the validity of the HPM. Theorists and researchers of health promotion should not assume that the causal mechanisms underlying health promotion behaviors in men and women are identical, but must forge ahead with an exploration and clarification of possible differences.