In this paper, we examine the effect of accounting conservatism on earnings persistence and the stock market's valuation of earnings. This issue is of particular interest to regulators as to whether and to what extent conservatism is desirable. In general, mismatching between revenues and expenses are more serious under more conservative accounting than under less conservative accounting, and such mismatching is likely to reduce earnings persistence. Thus, we hypothesize that more conservative earnings are less persistent than less conservative earnings. Moreover, Ohlson (1995) indicates that less persistent earnings obtain smaller pricing multiples than more persistent earnings do. Based upon these arguments, we further hypothesize that the pricing multiple on more conservative earnings is smaller than that on less conservative earnings. Empirical findings from U.S. Compustat companies during the period of 1984-2003 support our hypotheses. Our evidence raises the concerns about the reduced earnings predictability associated with conservatism.