We develop post-processing approaches based on linear regression that make ensemble forecasts more reliable. First of all we enforce climatological reliability (CR) in the sense that the total variability of the forecast is equal the variability of the observations. Second, we impose ensemble reliability (ER) such that the spread around the ensemble mean of the observation coincides with the one of the ensemble members. Since, generally, different ensembles have different sizes, standard post-processing methods tend to overcorrect ensembles with large spreads. By taking variable values of the error variances, our forecast becomes more reliable at short lead times as reflected by a flatter rank histogram. We illustrate our findings using the Lorenz 1963 model.