It has been argued that one of the functions of fringe benefits is to reduce turnover. However, due to a lack of data, the effect on quits of the marginal dollar of benefits relative to the marginal dollar of wages is an under-researched topic. This paper uses the benefit incidence data in the 1979 Cohort of the National Longitudinal Survey of Youth (NLSY79) and the cost information in the National Compensation Survey to impute benefit costs. The value of imputed benefits is then entered as an explanatory variable in a mobility equation that is estimated using turnover information in the NLSY. We find that the quit rate is much more responsive to fringe benefits than to wages; this is even more the case with total turnover. We also find that benefit costs are correlated with training provision. Due to the high correlation of the costs of individual benefits, it is not possible to disentangle the effects of separate benefits. An interesting feature of the model that we develop for interpreting the strong negative relationship between fringe benefits and turnover is that abstracting from heterogeneity, workers must at the margin place a higher valuation on a dollar of wages than a dollar of benefits since otherwise an employer could profit by switching compensation from wages to fringes. Worker heterogeneity modifies this result and reinforces any causal relationship between fringe benefits and turnover provided that more stable workers have a greater preference for compensation in the form of fringes.