Graph theoretic concepts were exploited to construct candidate computer models describing the hydroisomerization of n-hexadecane over bifunctional catalysts at a mechanistic level. The molecules and carbenium ions were represented as atomic connectivity matrices and the reactions as matrix operations. The computer-generated models were then automatically converted to a set of ordinary differential equations representing the reactions in a plug flow reactor. Typical diagnostics for the model building process indicate that models containing about 1000 species (molecules and ions) and 3000 reactions could be generated in around 1000 CPU seconds. Thus, some 25 candidate hydroisomerization models were built and evaluated through qualitative and quantitative comparisons with experimental data. The resulting best model was further pruned by removing extraneous reactions which did not result in observable products, This model was tuned against experimental data on three different catalysts at different temperatures. Good agreement between model predictions and experimental data was obtained. Trends in the fundamental mechanistic rate constants obtained from the model fitting process were in harmony with the trends in the acid and metal loadings of the catalysts. Thus, fundamental mechanistic modeling can provide valuable insight into the nature of the catalytic action.