In response to a need for more primary care physicians and patients' growing attraction to alternative health care, greater numbers of individuals are applying to naturopathic colleges. With increasing numbers of applicants, better methods of predicting potential effectiveness as an N.D. are needed. This study examined factors (both academic and psychosocial) that best predict success in naturopathic school. Demographic, academic, and psychosocial survey data were collected from thirty-three students who had just completed their second year of naturopathic medical school. This information was correlated with scores on the NPLEX Basic Science exams which were taken the following summer. Because of the small sample size, a bootstrap resampling technique was used to produce estimates for a hierarchical regression. Demographic variables (sex, age, whether or not English was the first language) and undergraduate major, explained almost 10% of the variance in Basic Science Exams (BSE) scores; however, none of these variables were significant predictors in the first step of the regression. As predicted, the addition of undergraduate grade point average (GPA) significantly increased the amount of variance accounted for (to 39.9%) in BSE scores. Also as predicted, adding the psychosocial variables to the model increased the amount of variance accounted for to 52%. This addition also made sex a significant predictor, but external locus of control was the only psychosocial variable which was significant in any of the models. The best model contained the psychosocial variables of both internal and external locus of control but not commitment and accounted for 51 % of the variance in BSE scores. Sex, undergraduate GPA, and external locus of control were significant predictors. Results are consistent with previous research using data on students from allopathic medical schools. However, complex relationships exist among the psychosocial variables and between the psychosocial variables and gender. The suppression effect of the psychosocial variables with gender, multicolinearity between the commitment and locus of control variables, and suppression due to common method variance between the internal and external locus of control variables are discussed. Limitations of bootstrap methodology are considered.