We report the results of a real-donation experiment in which we test the effect on giving to charity of tangible and intangible house and earned money. We also expand the action set, allowing our participants to take from, as well as give to, their charities. In our intangible treatments, the allocation decision is indicated on a standard decision sheet. In our tangible treatments, endowments are received as currency and participants indicate their allocation decisions by anonymously allocating their endowments between own and charity envelopes. We test if in a highly anonymous environment with cash in hand, will participants give to, or take from charity? We also test if participants’ behavior depends on the nature of their endowment: earned or house money? We report three results. First, tangible endowments are found to reduce, but not eliminate, charitable giving. Second, there is significantly more taking in the tangible treatments than the intangible treatments. Finally, we do not observe a consistent earned money effect. With tangible endowments, earned money reduce giving. For intangible endowments, we find that earned money increase giving. This is in contrast to prior studies that found that earning one’s endowment reduced giving in standard dictator games (i.e., those with anonymous other subjects as recipients). Our results suggest that the nature of the recipient, and possibly the action set, are important factors for giving decisions in the laboratory. Implications for experimental design are discussed.