The recent development of a detailed kinetics mechanism for isooctane from Lawrence Livermore National Laboratory (LLNL) has made the prediction of the ignition behavior of automotive fuels feasible. However, the large size of the mechanism translates to burdensome computational requirements especially when coupled to CFD for engine design purposes. We were able to greatly reduce the size of the LLNL isooctane mechanism from 857 species and 3606 reactions down to 63 species, while preserving its predictive capability in the HCCI regime. This remarkable size reduction was achieved by first reducing the detailed mechanism to a skeletal mechanism, then to a reduced mechanism by using quasi-steady state assumption (QSSA). A novel Targeted Search Algorithm (TSA) is developed to systematically screen species for QSSA in order to reduce the size of the reduced mechanism while maintaining accuracy. This new approach is found especially useful when the chemical kinetics involves complex ignition pathways. The 63-species reduced mechanism generated by TSA is tested extensively to ensure small overall errors induced by QSS species. Later in the paper, we present the results of the ignition delay time calculations, transient WMR calculations, emission predictions, and HCCI engine simulations with KIVA-3V with various mechanisms.