Logistic regression has emerged as a robust alternative to discriminant analysis. This paper compares both statistical procedures in terms of model specification, assessment of fit, and interpretation. Although both statistics performed well as classification techniques, the logistic regression model was more parsimonious and easier to interpret. The author concludes that when choosing between these two statistics, social work researchers should consider: (1) the purpose of the analysis (classification or description); (2) the characteristics of the sample; and (3) the tenability of assumptions.