This study theoretically and empirically examines altruistic and joy-of-giving motivations underlying contributions to charitable activities. The theoretical analysis shows that in an economy with an infinitely large number of donors, impurely altruistic preferences lead to either asymptotically zero or complete crowd-out. The paper then establishes conditions on preferences that are sufficient to yield zero crowd-out in the limit. These conditions are fairly weak and quite plausible. An empirical representation of the model is estimated using a new 198692 panel of donations and government funding from the United States to 125 international relief and development organizations. Besides directly linking sources of public and private support, the econometric analysis controls for unobserved institution-specific factors, institution-specific changes in leadership, year-to-year changes in need, and expenditures by related organizations. The estimates show little evidence of crowd-out from either direct public or related private sources. Thus, at the margin, donations to these organizations appear to be motivated solely by joy-of-giving preferences. In addition to addressing the basic question of motives behind charitable giving, the results help explain the existing disparity between econometric and experimental crowd-out estimates.