A situation is considered where a number of models can be proposed for a system : a chemical reaction, for instance. These models, based on a certain pattern of behaviour situation is considered where a number of models can be proposed for a system : a or mechanism, are claimed to be equally plausible for the given system. The model which most adequately describes the underlying phenomenon must, therefore, be chosen. A novel procedure is proposed for accomplishing this task. A versatile distance function is employed for designing experiments as well as for analysis of data for discrimination among the proposed models so as to establish the credibility of a single model. The potential of the procedure is demonstrated through its implementation to linear and nonlinear mechanistic models, comparing as well its performance with other procedures reported in the literature.