In this paper we present a forecasting method for time series using copula-based models for multivariate time series. We study how the performance of the predictions evolves when changing the strength of the different possible dependencies and compare it with a univariate version of our forecasting method introduced recently by Sokolinskiy & Van Dijk. Moreover, we also study the influence of the marginal distribution with the help of a new performance measure and lastly we look at the impact of the dependence structure on the predictions performance. We also give an example of practical implementation with financial data.