In this paper we examine a number of issues concerning the validation of two types of macro models: Aggregate response models, and aggregate flow models. The discussion is preceded by some remarks about model construction. The validation process is considered to consist of two steps: model fit and model prediction. Model fit will mostly involve econometric techniques as far as aggregate response models are concerned, tracking or trial-and-error-estimation for aggregate flow models. We will emphasise the need for good data and will argue for increasing the frequency of observations. The second step, model prediction, is in many cases not carried out by lack of data. A brief discussion is given of an empirical study about the relationship between market share and outlet share for the gasoline market in a European country.