From volumetric method to isotope dilution method, K-Ar dating has got its period of maturity. But traditional K-Ar dating techniques could not overcome the errors coming from the influence of isotope mass discrimination. It is possible to establish quantitative relationship between peak height and abundance of Ar. A new method of K-Ar dating with peak comparison method has been produced, in which the spike (38Ar) has not been used again. With this method, the errors coming from mass discrimination could be eliminated. This is especially important to young volcanic rocks dating. Recent volcanic substances contain initial argon, which underwent mass fractionation in some extent, and their isotopic ratios fell on the mass fractionation line. This implies that the fractionation influence the KAr dating of very young samples. According to accurate measurement of 38Ar/36Ar ratio of samples, initial 40Ar can be precisely subtracted from the total Ar and the radiogenic 40Ar obtained. This method can be used in dating old rock and mineral samples, with their results corresponding well with that of K-Ar dilute method and 40Ar-39Ar method, and more importantly, used in very young samples. This method can be used to date samples as young as 2 ~ 5ka, where the error is less than 1ka for high potassium samples and the relative error is not more than 1.5% for samples older than 0.1 Ma. This paper introduces the basic principles and analysis process of peak comparison method and at the same time the requirements and range of application are also discussed.