This paper examines the predictive ability of financial ratios, specifically dividend yield (DY), book-to-market (B/M), and earnings-price ratio (E/P), in forecasting stock returns. The author focuses on monthly returns and uses a new method that incorporates the autocorrelation of the predictive variable to correct for small-sample biases. The results show that DY has strong predictive power, with a significant relationship between DY and returns from 1946 to 2000, both in the full sample and in subperiods. B/M and E/P also exhibit some predictive power, but it is weaker and less consistent. The paper highlights the importance of considering the autocorrelation of the predictive variable, which can significantly enhance the power of empirical tests. The findings suggest that financial ratios can be useful tools for predicting stock returns, despite previous studies suggesting otherwise.This paper examines the predictive ability of financial ratios, specifically dividend yield (DY), book-to-market (B/M), and earnings-price ratio (E/P), in forecasting stock returns. The author focuses on monthly returns and uses a new method that incorporates the autocorrelation of the predictive variable to correct for small-sample biases. The results show that DY has strong predictive power, with a significant relationship between DY and returns from 1946 to 2000, both in the full sample and in subperiods. B/M and E/P also exhibit some predictive power, but it is weaker and less consistent. The paper highlights the importance of considering the autocorrelation of the predictive variable, which can significantly enhance the power of empirical tests. The findings suggest that financial ratios can be useful tools for predicting stock returns, despite previous studies suggesting otherwise.