||The thesis reviews the selected topics of application of nonlinear models and ideas of chaos theory in economics, particularly in capital markets theory. Efficient markets hypothesis was the leading concept on this field for long time supported by dozen empirical studies. At the beginning the stylized facts of return time series are resumed. Then we describe basic concepts and methods of chaos theory and review present attempts to detect a chaotic dynamics in a stock price series. The results of these tests often find strong evidence for nonlinear dependence, but no convincing evidence for chaos. The BDS test is described in detail. It tests the null hypothesis that the random variables are independent and identically distributed against unspecified alternative. It can detect various types of dependence, both linear and nonlinear. Some other types of nonlinear dependence (e.g. conditional heteroscedasticity) that can be present in financial series are concidered. Some of the methods and the tests described before are used to analyze the daily returns of the stock index S&P 500 in period of 1964-1992. Two new theories (coherent market hypothesis and fractal market hypothesis) that reflect recent findings of the stock market returns research and may be considered as an alternative to efficient markets hypothesis are introduced. At the conclusion we suggest possible directions of further research in this area and how its results might be utilized by practicioners.