We use controlled experiments to identify the proximal causes of gender differences in the repayment of microcredit. We recruit male and female subjects from a patriarchal and a matrilineal community in Bangladesh, who live in the same villages, and find that the female subjects have a greater willingness to repay microcredit in every society irrespective of the type of loan. Thus, the observed gender differences in the repayment of microcredit cannot be explained by the different roles that women play in different societies. In other words, women are “naturally” better at credit risks than men in microcredit. We confirm that our results are not driven by the common culture and values among our subjects that stem from geographical proximity. Employing the microfinance repayment game allows us to investigate the drivers of a borrower’s willingness to repay a microloan. In our modified version of the game, we extend small loans to the subjects under individual and joint liability-based loan contracts, with the provision of follow-up loans upon successful repayment of a current loan, i.e., we include a dynamic repayment incentive. Each subject independently invests in a risky project. If the project fails, the subject cannot make any repayment. The origin of gender differences in human behavior has been a long-debated issue.