Empirical researches in financial literature have shown evidence of a skewness and a time conditioning in the univariate behaviour of stock returns and, overall, in their dependence structure. The inadequacy of the elliptical and, in general, symmetrical multivariate constant model assumptions, when this type of dependence occurs, is an almost stylized fact. Beyond these characteristics, recent studies have highlighted a dynamic whole/tail dependence which changes over time. This paper provides a new approach for modeling multivariate financial asset returns based on time-varying vine copulas. A dynamic multivariate model is proposed, based on a decomposition of d-dimensional density into a cascade of bivariate conditional and unconditional copulas, with two desirable features: a flexible construction, which allows for the free specification of d(d −1)/2 copulas and easy computation in presence of high dimensional data.