The B. F. Goodrich Company spends considerable money selecting and training territory managers. The hiring of territory managers generally rests with local management's "intuition." Since 1967, a substantial percentage of these highly trained salesmen have separated from the company. It was within this attrition context, that an exploratory, objective, selection program was studied. A multiple discriminant function (multigroup model) was used to analyze three groups of B. F. Goodrich Territory Managers' California Psychology Inventory and Edwards Personal Preference Schedule test scores. The data revealed that known territory managers' group membership could be significantly predicted. A longitudinal study (one year later) showed a high percentage of accuracy (i.e. 82 percent) between predicted test score probability group membership and original group membership. Further the multiple discriminant function analysis of territory managers' test scores disclosed that only a few test variables need to be considered in making a probability hiring decision.