This dissertation consisted of an empirical analysis of the determinants or initial occupational choice by male high school graduates. The approach used was based on the theory of random utility. According to this approach, the individual selects a particular outcome from a set of possible outcomes based on both observed and unobserved characteristics of the individual and the particular possible outcome. In this analysis, the occupational choice set contained three possible outcomes. These possibilities were civilian sector employment, military service and college enrollment. For empirical analysis, a sample of 1,748 male high school graduates was drawn from the National Longitudinal Survey of Youths (1979-1981). The empirical model consisted of a mixed discrete/continuous simultaneous 4 equation system. Three estimation strategies were used. The first was a sample two stage logit/ordinary least squares procedure. The second was a modified two stage logit/ordinary least squares procedure that corrected for self-selectivity bias. the third strategy consisted of a modified two stage logit/ordinary least squares procedure that corrected for both self-selectivity and choice-based sampling bias. The estimation results indicate that the decision to enlist is most sensitive to the net income of the individual's family and the predicted civilian sector wage. The military experience of the individual's father and the desire to acquire additional training are also important in this decision. In addition, the differences in the estimates across the three estimation procedures illustrate the importance of correcting for sample biases.