In this report results of applying time series models for assessing the thermal performance of the IEA Annex 58 test box based on data given in the Common Exercise 4 (CE4), which was measured in Almeria, Spain. Both ARX, ARMAX and grey-box models are applied. Finally, the same models are fitted for the Common Exercise 3b (CE3) data measured in Belgium and the results are compared. The focus in this report is on model selection and validation enabling a stable and reliable performance assessment. Basically, the challenge is to find a procedure for each type of model, which can give un-biased and accurate estimates of the essential performance parameters, including reliable uncertainties of the estimates. Important is also the development of methodologies for analyzing the quality of data, for example correlated inputs and lack of information in data (e.g. if no clearsky days with direct solar radiation is present data), these aspects are discussed. Furthermore, new models for enhancing the description of the effect of solar radiation on the test box is presented.