In this study, the current literature on volatility prediction in financial electricity markets is extended by incorporating volatility spillover and covariance effects. This is done using the concept of realized volatility and covariance with three financial electricity contracts traded at the Nord Pool exchange. A recently developed robust method to separate the total variation into continuous and jump components is applied. The results indicate that volatility spillover and covariance effects are clearly present in the three different contracts. The prediction accuracy is improved significantly when utilizing a multivariate framework (vector autoregressions) rather than a univariate one.