In this research, I classify all stocks into two groups: dividend and non-dividend payers and hypothesize that profitability premium may only exist among the firms with unforeseeable future cash flows, i.e., non-dividend payers. As expected, my empirical results support that profitability premium only exists among non-dividend payers but is very trivial among dividend payers. Dividends have a moderate effect on profitability premiums. To dig further into the source of profitability premium, I investigated risk and behavioral explanations from three perspectives: macroeconomics, industry, and total risks investors perceive for a firm. The evidence from empirical analysis supports that the profitability premium is mainly driven by the overpriced, unprofitable non-dividend payers, which, on average, have negative earnings announcement returns. In contrast, there is no significant positive or negative abnormal return from earnings announcements for portfolios sorted by profitability among dividend payers. Furthermore, the evidence from analyst forecast errors confirms that analysts are over-optimistic about unprofitable non-dividend-paying stocks and disagree more with their EPS forecast. Overall, the study finds that investors' expectation errors are the source of the profitability premium. It rejects the idea that risk is the profitability premium driver.