Ensemble-based methods have been successfully applied in reservoir history-matching problems in the last decade. Among the advantages normally attributed to these methods, the fact that they generate multiple realizations of the model is one of the most important. By simulating these realizations, one can estimate the uncertainty in the production forecast for the field. However, because of limitations related to the use of relatively small ensembles, these methods often underestimate the posterior variance in the reservoir model parameters. Consequently, they tend to underestimate uncertainty in production forecasts. This paper introduces a simple procedure to evaluate the uncertainty bounds in the field production using ensemble-based data assimilation. The implementation of the proposed method is straightforward requiring very few modifications in a standard data assimilation code. The method was tested against the PUNQ-S3 case and a real field problem.