||This thesis should contribute to research in the area of fundamental analysis. Its aim is to study whether financial statement data of Czech non-financial companies capture information that is not reflected in prices. Therefore, the question is whether investment strategy based on financial statement analysis could earn excess returns. In order to test this hypothesis, a three-step estimation procedure based on a logit model is used to identify financial ratios relevant for prediction of future earnings. The final estimated model includes four financial ratios and is then used to set a one-year investment strategy. Although the performance of the estimated model is not too sound, this investment strategy brings positive abnormal returns during the monitored period of time. Despite the fact that results were influenced by several factors, they could indicate that financial statement analysis of companies listed on the Prague Stock Exchange is able to predict stock returns.