We analyze the effects of non-synchronicity and market microstructure noise on realized covariance type estimators. It is shown that non-synchronicity leads to severe biases, whenever synchronization methods that employ last-tick interpolation are used. We study a simple estimator which resolves...