The invention discloses an engine management system (EMS) oil product learning method, which comprises the following steps of: judging whether a first-class oil product self-learning value A of the EMS meets a rapid correction condition; if the first-class oil product self-learning value A does not meet the rapid correction condition, executing slow correction of the first-class oil product self-learning value A to obtain the slowly corrected first-class oil product self-learning value, wherein the slow correction is to change one-way correction of the original system into double-way correction and add a correction condition, and integrator input of the first-class oil product self-learning value A is determined according to air-fuel ratio closed-loop control deviation; and if the first-class oil product self-learning value A meets the rapid correction condition, executing rapid correction to obtain the rapidly corrected first-class oil product self-learning value A by taking the difference value of a second-class oil product self-learning value B and the first-class oil product self-learning value A as integrator calculation input. By the EMS oil product learning method provided by the invention, oil products can be identified accurately, oil product learning deviation is avoided and oil product learning accuracy is improved.