An off-line algorithm for empirical modeling and identification of non-linear dynamic systems is presented. The minimal input to the algorithm is a sequence of empirical data and the model order. Using this information, the algorithm searches for an optimal model structure and parameters within a rich non-linear model set. The model representation is based on the interpolation of a number of simple local models, where each local model has a limited range of validity, but the local models yield a complete global model when interpolated. The method is illustrated using simulated data.