In a study of turnover in a midwest trucking firm, the files of 199 terminated drivers were used in a statistical analysis of 19 variables that might have been correlated to length of service. A series of interviews with drivers and key management personnel supplemented the statistical data. It was found that previously identified biographical correlates were less predictive than expected. It was suggested that these findings resulted from intervening variables, such as an adverse working environment, dissatisfaction, and perceived high opportunity. These results highlighted some of the logical problems with identifying intervening variables, the need for an alternative method of statistical analysis, and the possibility of a "satisfaction threshold" below which traditional correlates lose their predictive ability. In short, the study implied that information about the present organizational climate and satisfaction with working conditions were valuable in the selection of correlates to turnover, the interpretation of statistical information, and the identification of possible intervening variables. The study also suggested important redefinitions of key concepts, such as satisfaction, opportunity, and length of service. (RL)