 Independence of uncorrelated jointly normal variables

Two subvectors of a jointly multivariate normal distribution are independent exactly when their cross-covariance matrix is zero.