Localization is an essential element of ensemble-based Kalman filters in large-scale systems. Two localization methods are commonly used: covariance localization and domain localization. The former applies a localizing weight to the forecast covariance matrix, while the latter splits the assimilation into local regions in which independent assimilation updates are performed. The domain localization is usually combined with observation localization, which is a weighting of the observation-error covariance matrix, resulting in a similar localization effect to that of covariance localized filters. It is shown that the use of the same localization function in covariance localization and observation localization results in distinct effective localization length-scales in the Kalman gain. In order to improve the performance of observation localization, a regulated localization scheme is introduced. Twin experiments with the Lorenz-96 model demonstrate that the regulated localization can lead to a significant reduction of estimation errors as well as increased stability of the assimilation process. © 2011 Royal Meteorological Society.