A major thrust in combustion research is the development of chemical kinetic models for computational analysis of various combustion processes. Significant deviations can be seen when comparing predictions of these models against experimentally determined combustion properties over a wide range of operating conditions and mixture strengths. However, these deviations vary from one model to another. It would be insightful in such circumstances to elucidate the species and subchemistry models which lead to the varying prediction ability in various models. In this work, we apply the alternate species elimination (ASE) method to selected mechanisms in order to analyze their predictive ability with respect to propane and syngas combustion. ASE is applied to a homogeneous reactor undergoing ignition. The ranked species of each model are compared based on their normalized changes. We further provide skeletal versions of the various models for propane and syngas combustion analysis. It is observed that this approach provides an easy way to determine the chemical species which are central to the predictive performance of a model in their order of importance. It also provides a direct way to compare the relative importance of chemical species in the models under consideration. Further development and in-depth analysis could provide more information and guidance for model improvement.