To teach global software engineering, we devised a complementary distributed module with a shared project involving both local and international teams. In local teams, students are located at the same university and trained in one of the two complementary topics. In international teams, students are located at two different universities and trained in one of the two complementary topics. This study empirically investigates whether the students in the international teams can compensate the extra effort required to deal with communication, coordination, and collaboration issues that characterize global software engineering projects with learning by osmosis (i.e., by transferring knowledge among globally distributed teams trained on different topics). The results show that there was no statistically significant difference between the performance of local and international teams.We assert that the students in the international and local teams perform equally well, thanks to learning by osmosis. However, our analysis of the self-reported questionnaire data revealed that most of the participants (i.e., 70%) would like to work in local teams in real-life project, 74% of the participants thought international teams were less efficient, and 41% of the participants reported lack of trust in their international team members compared with their local team members. Copyright © 2011 John Wiley & Sons, Ltd.