In this paper we investigate how schoolmates influence high school dropout intentions in Catalonia, Spain. Our analysis uses self-reported friends to identify possible peers by assuming that peer influence flows in one direction in cases where one student identifies another as a friend, but the other does not reciprocate. We first estimate the effects of education aspirations of non-reciprocating friends on students' own education aspirations, with and without conditioning on a large set of personality and cognitive characteristics. We then investigate the extent to which the estimated effects are associated with friends' height, weight, BMI, gender and cognitive ability. The estimated impact of non-reciprocating peers' dropout intentions is small and generally not statistically significant: a 10 percentage point increase in the fraction of non-reciprocating peers that intend to drop out increases students' chances of dropping out by about 0.2 percentage points. (Contains 3 tables and 1 figure.)