OUR UNDERSTANDING of why corporations pay dividends is currently unsatisfactory. On the one hand, received theory tells us that dividends are irrelevant (in the sense that any two arbitrarily chosen dividend policies have equivalent consequences), both in the absence of taxes (Miller and Modigliani [13]) and in their presence (Miller and Scholes [14]). On the other hand, dividends continue to flood the empirical world with cash as regularly and as consistently as the sun scorches the desert, and one is hard put to characterize this pattern (currently at an annual rate of about $63 billion') as being founded on irrelevance. Not surprisingly, the attendant anomaly has led some, notably Black [1], to suggest that we really don't know why companies pay dividends. Something is clearly amiss. The present paper will look to the information content of dividends as a substantive (although not necessarily complete) explanation for the prevalence and persistence of positive dividend policies in market economies. The notion that dividends may constitute a source of information is, of course, not new (see e.g., Miller and Modigliani [13], Black [1], Stern [18]). The basic idea is that the raising and lowering of dividends communicates information over and beyond what is provided by (mandated and nonmandated historical cost-based) earnings reports, forecasts, and other announcements. By in effect merging extant dividend theory with the theory of public information, the present paper generates several noteworthy consequences. In particular, it extends substantially the current paradigm, bringing within its domain cases in which dividend payments are