The models which are in use in the financial world are based on the following assumptions: The continuous daily returns stick to joint normality and the i.i.d. (identically and independent distribution) concept. Furthermore all moments have to be constant all the time, especially the variance (homoscedasticity). Finally, the dependencies between the assets are measured by the correlation which is also assumed to be constant all the time. The target of the time series analysis with the selected real estate returns is to examine the assumptions regarding its validity. The conclusion of the study is that the assumptions are untenable, especially in crisis, when the models are requested to be most reliable.The focus of this master thesis is to deal with two subject areas, namely ?Real Estate? and ?Risk Management?, together. At the beginning of the qualitative fraction the following parts are described: The market of the real estate products, the real estate portfolio management with the principles of diversification and some theoretical aspects of Risk Management, for example the types of risk and the processes of the risk management.The following empirical study is conducted with the programming language ?R?. First of all there is a calculation of the third and forth moment of the distribution (Skewness, Kurtosis), the Jarque-Bera-Test of joint normality and the autocorrelation of the daily real estate returns followed by an application of the most important methods of Risk Management: Volatility, Maximum Drawdown, Value-at-Risk plus Cornish-Fisher, Correlation corresponding to Pearson?s and Spearman?s definitions and last but not least Lower-Tail-Dependence. For the Correlation and the Tail Dependence a rolling method is also applied. The quantitative part concludes with three individual real estate portfolios described and calculated: The Minimum-Variance-Portfolio (MVP), the Tangency Portfolio and the Constrained MVP.