In this paper the authors investigate the predictors of long-term care placement. Regression analysis is used to estimate a multivariate placement model. Variables found to be significantly related to placement are sex, marital status, help from relatives, client and family preferences, index of ADL, number of medical conditions, ability to take medications, ability to make decisions, and income. The estimated function explained 68 percent of the variance. Discriminant analysis and logistic regressions were used to evaluate the performance of the estimated placement function. The authors stress the importance of learning more about the placement process before making causal inferences about the cost-effectiveness of alternative long term care settings.