Many important activities, such as charitable giving, voting, and paying taxes, are difficult to explain by the narrow self-interest hypothesis. This paper tests conditional cooperation in a field experiment. The field experiment about charitable giving supports the theory of conditional cooperation: contributions increase, on average, if people know that many others contribute. The effect varies, however, depending on past contribution behavior - those who never contributed do not change their behavior, while people who are indifferent about contributions react most strongly to information about others' behavior. Section I presents the field experiment and the empirical strategy to test the hypotheses, Section II shows the results, and Section III offers concluding remarks.