Financial statement analysis is a method that seeks to predict stock returns by extracting factors that affect firm value from publicly disclosed financial statements. This study raises the question of whether predicting stock returns using financial statements remains useful when cash flows—a key type of information provided by financial statements—are considered as a performance measure, in addition to accounting earnings, which have traditionally served as a proxy for firm value in financial statement analysis. This study focused on refining the methodology by employing buy-and-hold returns and equally weighted index returns, and by considering market-adjusted returns (MAR) and size-adjusted returns (SAR) as measures of abnormal returns. A randomization strategy was used to calculate the significance levels of hedge portfolios. The sample consisted of 159 firms listed on the Korea Stock Exchange as of December 31, 1980, with the first five-year period from 1982 to 1986 used for logit model estimation, and the subsequent six-year period from 1987 to 1992 used for testing the model's predictive power and analyzing returns. This study hypothesized that if future earnings (future cash flows) one period ahead can be predicted using financial statements, then stock returns should also be predictable using such predictive power, and empirically tested this hypothesis. Forty-five ratio variables were used as financial statement information, and the prediction model for the sign of abnormal returns was estimated using data from the estimation period and logit models. Next, portfolios were constructed based on the probability (Pr) of positive abnormal returns calculated from the prediction model according to established probability criteria, and it was tested whether abnormal returns were realized after a one-year holding period. The main results of this study can be summarized as follows. The investment strategies using both the future earnings model and the future cash flows model did not yield significantly positive returns in terms of MAR, SAR, or Jensen's alpha controlling for size and risk. The combined model integrating cash flows and earnings also did not produce positive abnormal returns.