This contribution deals with a simulation analysis of time varying correlations. Several different approaches are assumed: the exponential smoothing, the moving averages, the technique based on the Cholesky decomposition with various types of estimators and Engle's dynamic conditional correlations models.