This paper formulates explicitly a life—time optimization problem that a person faces at each period t and suggests two methods of estimating the underlying parameters of the preferences and the constraints. We focus on one discrete choice variable such as fertility, schooling and labor—force participation. An example for each model is provided where the current decision depends on fast choice variables. In order to estimate this type of a model it is necessary to solve for the state dependent conditional probabilities for the discrete choice at each t. Two ways for calculating these conditional probabilities are provided. The first is calculated from the full solution for the dynamic programming problem. The second is found by using some necessary conditions for the optimal solution of the individual. We compare the two methods by a Monte—Carlo example using the fertility model. The full solution metnod is used as the "true" model and the other method generates acceptable values for the parameters and the likelihood function.