The existing hypothesis of dynamic stock adjustment is modified to allow for discontinuous adjustment observed in the actual behavior of the household toward durable goods purchases. It is proposed that the probability of purchase of durables is determined by the gap between desired and actual stock; and, given purchase, the amount of purchase is also determined by the gap between desired and actual stock. Difficulties of estimating such a model from cross-section data are indicated and a reformulation of the model is suggested. Regression equations are then specified and estimated from Survey of Consumer Finances data.