Both model identification and prediction are central to the practicing ecologists tool set. Correct identification is necessary for the best possible understanding of the system under consideration and for the best possible management decisions. ICOMP is a potent new technique that will be of great use to scientists in our field. The ICOMP criterion contains a penalty term based on the accuracy of parameter estimates and interactions between parameters. This work extends the validated realm of application of the ICOMP model selection criterion beyond that documented in the statistics literature. These previous studies have focused on how well ICOMP behaves in cases of multivariate linear regression; we have extended this work to the case of nonlinear models. Traditional tools may not be adequate and examination of new tools is crucial for the discipline of ecology as models become more complex. Results of this study indicate that the ICOMP criterion's performance is superior to that of the traditionally used AIC, AICc and BIC when selecting from a candidate model set including complex nonlinear models.