We use information on students’ past participation in economic experiments, as stored in our database, to analyze whether behavior in public goods games is affected by experience (i.e., previous participation in social dilemma-type experiments) and history (i.e., participation in experiments of a different class than the social dilemma). We have three main results. First, at the aggregate level, the amount subjects contribute and expect others to contribute decrease with experience. Second, a mixture model reveals that the proportion of unconditional cooperators decreases with experience, while that of selfish individuals increases. Finally, history also influences behavior, although to a lesser extent than experience. Our findings have important methodological implications for researchers, who are urged to control for subjects’ experience and history in their experiments if they want to improve the external validity and replicability of their results.