When a thermoelectric generation (TEG) system is used to recover waste heat from gasoline engine exhaust, the thermal parameters of the exhaust gas vary greatly; this has an influence on the optimal performance of the TEG system. To improve the recovery of exhaust waste heat and its conversion to electric power effectively, the peak power evaluation and optimal performance analysis are conducted on a TEG system with different exhaust thermal parameters. A sandwich plate-type exhaust heat exchanger is modeled using finite element analysis. The maximum net power is the parameter chosen for optimization, which is achieved by identifying the optimal length, width, and height. The results show that the flow velocity and TEG module area are two key parameters for optimization of the TEG system. Furthermore, all the optimal performance parameters are derived by deducing correlation equations. Using these correlations, the optimal performance and ideal peak net power of the TEG system can be conveniently obtained for any given exhaust thermal parameters. This is an effective method to determine the optimal design dimensions and evaluate the net power recovery ability of the designed TEG system.