Standardized patch tests were performed on 3,974 patients between August 1, 1985 and July 31, 1989. Significant determinants of positive responses based on previously reported bivariate analysis of response were entered into a logistic regression model; the contribution of each factor to response to each of 27 test substances with which the patients were tested was evaluated. For 14 test substances, age increased the odds of response; nickel was striking in that increasing age decreased the odds of a positive response. Specific contactants were found to be associated with gender and topographical location of the eczematous dermatitis leading to the application of the patch tests. Atopic dermatitis was associated with a lower odds of response toP-phenylenediamine, but in no instance was associated with increased odds of a positive response.