Producing an oil well is a transient process, and although we like to assume there is a steady state phase in the production life, it is a fragile situation. Due to the many unknowns in the reservoir, changes in well production performance cannot easily be predicted. Changes in the subsurface will influence the production and lifetime of the well and need to be identified as soon as possible. Consequently, monitoring well performance is of great importance. Identifying deviations from production forecasts is difficult, because mismatches between predicted and measured flow rates can have two causes: either the production prediction model is not valid, or the production test data is wrong. This research aims to quantify the uncertainty inherent to performance models caused by the uncertainty in input parameters and by the method used to establish the model. This is done by introducing the performance envelope, a range instead of a line in which the well is expected to produce for the given input parameters and the used method to establish it. Two workflows are proposed to establish such a performance envelope, one for fitted models and one for simulated models. Additionally, a rate selection method is proposed, optimizing the test program by selecting the three rates which result in the narrowest confidence interval. The results show that a Performance Envelope can be computed for both fitted and simulated performance models, but that further research is needed to use them as an objective criterion. Furthermore, they show that rate selection has influence on the quality of the resulting model and that it can be used to optimize a test program.