Recently, much research has been carried out on secondary O2 feedback which performs control based on the output from a secondary O2 sensor (HEGO sensor).In this research it has been found that, regardless of catalyst aging conditions, the HEGO sensor output indicates 0.6 V when the catalyst reduction rate is maintained at the optimum level. Therefore, based on this relationship, we designed an accurate secondary O2 feedback with the aim of reducing emissions by stabilizing the HEGO sensor output to 0.6 V.In order to realize this control, it was necessary to solve the three problems of nonlinear catalyst characteristics, dead time characteristics, and changes in dynamic characteristics due to catalyst aging conditions. Therefore, these problems were solved using the modeling approach of robust control and a new robust adaptive control named Prediction and Identification Type Sliding Mode Control. This new control is comprised of an adaptive sliding-mode controller, predictor and identifier.As a result, we were able to increase the relative frequency of the HEGO sensor output in the range of 0.5 to 0.7 V to approximately 80%, regardless of the catalyst aging conditions. This in turn reduced NMHC emissions by about 30% and NOx emissions by about 40% over the FTP.