In many African countries the efficiency of public services deteriorated after independence as governments hired too many employees, allowed earnings to erode and performance standards to decline. Various explanations have been offered for this. Some have focused on the state’s role as an employer of last resort of graduates from domestic colleges and universities and its effects on the payroll. Others view public employment as an instrument of patronage, arguing that it was used to reward particular ethnic groups or regions for their political support. Using a binary logistic model this paper analyses the effect of merit-based criteria (education, age/experience) and ascriptive criteria (ethnicity or region of origin) on the probability of holding a public sector job in Kenya, Uganda and Tanzania. It finds that educational level, age and the developmental level of a respondent’s place of birth have a large influence on an individual’s likelihood to hold public sector employment, while ethnic identity has only a minor effect once other factors are controlled for. The findings support the first proposition that the state was a default employer of highly educated workers in the decades of independence and politicians thus exercised relatively little discretion over the allocation of skilled jobs. Moreover, graduates from peripheral and less developed regions of their respective countries were more likely to enter public employment than their counterparts from prosperous regions, suggesting that graduates from ethnically ‘advantaged’ backgrounds may in fact have a preference for private rather than public sector careers.