When information about the true abilities of job-seekers and applicants are hard to get, statistical discrimination by employers can be an efficient strategy in the hiring and wage setting process. But statistical discrimination can induce costs, if labor relations cannot be terminated in the short term and wages are fixed over a certain period. In this paper we use a unique longitudinal survey that follows the PISA 2000 students in their educational and work-life career. We test whether deviance in the PISA test scores from what one would have predicted based on observable characteristics, influences the probability to succeed in the transition from compulsory school into a firm-based apprenticeship and whether it can explain differences of the individual performances during training. Our results suggest that hard-to-get information plays a significant role in the transition, but not always in a symmetric manner.