Aim of this work is to evaluate the overall effect of social origins on secondary school track enrolment in Italy, Germany and Netherlands, allowing for consistent cross country_x000d_ comparisons. PISA 2003 is employed. Track choices are assumed to depend on student's ability and social origins; since proficiency before tracking is not observed, ability is not kept under control. Nonetheless, the unconditional social background effect is the quantity of main substantive interest because it represents the total effect of social origins on school choices. Yet, since regression coefficients in logit models are biased even with independent unobserved heterogeneity, comparison across countries are difficult; the average sample derivative of the response probability is employed instead and it is showed to be a valid alternative measure of the total social origins effect. The following issue is also addressed: social origins inequality in secondary school choices may be affected by access restrictions policies, at work in some countries, where enrolment into the more prestigious tracks is subject to binding school recommendations or ability tests. First, we propose a simple theoretical model and we derive that the policy is expected to lower the effect of social origins conditional on ability, although the impact on the total effect can either decrease or increase. Second, by exploiting the institutional differences across German Länder with respect to enrolment policies, we carry out a preliminary empirical analysis within Germany. The main empirical findings are: (i) the total effect of social origins on track choice is weaker in the Netherlands and stronger in Germany, with Italy in between; (ii) within Germany, access restriction seem to weaken the parental background effect._x000d_ _x000d_ Keywords: inequality of opportunity, school tracks, access restrictions, unobserved heterogeneity, logistic regression.