© 2019 Elsevier B.V. We study intertemporal crowding between two fundraising campaigns for the same charitable organization by manipulating donors’ beliefs about the likelihood of future campaigns in two subsequent field experiments. The data shows that initial giving is decreasing in the likelihood of a future campaign while subsequent giving increases in initial giving. While this refutes the predictions of a simple expected utility model, the pattern is in line with a model that allows for (anticipated or unanticipated) habit formation provided that donations in the two periods are substitutes.