This paper discusses some implications for the study of marital stability of a causal model in which education and occupation are seen as causes of income. Proper time order then places marital stability events after all three of these measures of social class. The class variable closest to instability is income. Therefore, assessment of the direct effects of education or occupation on marital stability must first remove the effects of income. Data from the 1960 Census using the percent of ever-married males living with the first spouse as a measure of marital stability are subjected to multivariate analysis. Little positive effect of education or occupation is found when the effects of income on stability are removed. Two alternative models which might explain how income acts to alter the risks of marital stability are discussed. A second measure of marriage behavior among ever-married males is analyzed. This study allows a reexamination of inferences that are often drawn regarding alleged white and nonwhite differences in attitudes toward marriage. An explanation of white and nonwhite differences in stability of first marriages is advanced. The possibility that income maintenance programs might increase marital stability is examined.