We investigate the effects of pay comparison information (i.e. information about what co-workers earn) and effort comparison information (information about how co-workers perform) in experimental firms composed of one employer and two employees. Exposure to pay comparison information in isolation from effort comparison information does not appear to affect reciprocity toward employers: in this case own wage is a powerful determinant of own effort, but co-worker wages have no effect. By contrast, we find that exposure to both pieces of social information systematically influences employees’ reciprocity. A generous wage offer is virtually ineffective if an employee is matched with a lazy co-worker who is also paid generously: in such circumstances the employee tends to expend low effort irrespective of her own wage. Reciprocity is more pronounced when the co-worker is hard-working, as effort is strongly and positively related to own wage in this case. Reciprocity is also pronounced when the employer pays unequal wages to the employees: in this case the co-worker’s effort decision is disregarded and effort decisions are again strongly and positively related to own wage. On average exposure to social information weakens reciprocity, though we find substantial heterogeneity in responses across individuals, and find that sometimes social information has beneficial effects. We suggest that group composition may be an important tool for harnessing the positive effects of social comparison processes.