Recent studies contend that answers to questions in the Survey of Consumer Finances reveal whether or not households are credit constrained. In these studies, households are defined as constrained if they applied for credit unsuccessfully or were discouraged from applying by the prospect of refusal. Households that applied successfully are classified as unconstrained. This approach is problematic in that it presumes that households do not take into account the presence and extent of credit constraints when formulating their loan demands. Credit constraints could have a more pervasive effect than precluding households from applying successfully. Anticipation of such constraints can also affect the amount of credit that households request. Successful applicants, therefore, need not attain the level of consumption that would be optimal in the absence of encountering a current borrowing constraint. Furthermore, unsuccessful applicants for loans comprise only a fraction of constrained households, namely those that overestimated the size of the loans likely to be granted. This dissertation uses a simple model of intertemporal optimization to demonstrate the effects of borrowing constraints on the demand for credit. It estimates a model of total debt of the households that applied for credit successfully. A key finding is that the demand for debt by particular groups that are likely to be constrained depends positively on current income. This is consistent with the presence of binding constraints among households that previous studies classified as unconstrained. Furthermore, this dissertation estimates a probit model of the outcome of the credit application. The results are interpreted in light of the hypothesis that unsuccessful loan applications primarily represent a forecasting error by households. This allows one to explain the lack of significance or apparent inconsistency of the coefficient estimates of certain variables.