Although government and companies invest in training programs, they seldom assess the impact of it. This work presents an evaluation study of a public training program. Hamblin's model of evaluating training programs (1978) was used to give theoretical support to this study. A quase-experimental research design was used containing three groups: two experimental groups and a control group. The duration of the study was two years with data collection before and after treatment. The impact on job behavior was assessed by evaluation instruments using four sources of information: the trained person, the supervisor, a colleague and a client, resulting in 2.468 questionnaires. Organizational changes and final value were assessed by hard indexes; the rise in numbers of professional education workshops and of apprentices (showing organizational changes) and the rise in numbers of employed persons (showing final value). Inspection of possible alternative factors that might explain the training impact, the GNP (Gross National Product), the interest's basic tax, unemployment tax and the changes in the legislation showed that, without treatment, the results would be opposite to those obtained. Hence, it might be concluded that the training programs indicating real benefits from the social investment.