Many European countries have delayed the time when school tracking starts in order to pursue equality of opportunity. What are the efficiency costs of de-tracking secondary schools? This paper builds a stylized model of the optimal time of tracking, estimates the relevant parameters using micro data for 11 European countries and computes the efficiency loss of tracking earlier or later than the efficient time. We find that this loss is moderate, and equal on average to half percent of GDP. (Contains 4 tables, 4 figures and 17 notes.)