We study a model that relates dividend payout policy to behavioral issues based on the ideas of mental accounting. A panel analysis across 29 countries and over 43,000 firm-years demonstrates that our model studying the relation between dividends and patience, loss aversion, and ambiguity aversion can be verified empirically. Our paper seems to be the first that highlights empirically in a straightforward way the relevance of behavioral patterns as important determinants for corporate dividend policy, while previous empirical studies could tackle this issue only indirectly. With several robustness tests we also address potential doubts concerning the quality of our data and analyze further implications of our theory.