This work studies dependence of extreme events between random variables and its implications in insurance Solvency II directive. Crucial concept is tail copula, what can be described as a tail analogy of general copula concept.Tail copulas are used to detect presence of tail dependence and they also determine its shape and stregth. Impact of tail dependence on value-at-risk of a portfolio is studied on a stylized example from insurance environment and results are compared with a method prescribed in Solvency II, which models dependence via classic concept of correlation.