The invention discloses a snoring sound identification method. The snoring sound identification method comprises the following steps: continuously acquiring n sound samples, recording duration times of all of the sound samples as T1, T2...Tn, and recording internal times between adjacent sound samples as t1, t2...tn-1, wherein n is larger than 1 and smaller than 100; judging whether the duration times T1, T2...Tn are in a first threshold range or not, and if yes, performing the next step; judging whether the interval times t1, t2...tn-1 are in a second threshold range or not, and if yes, performing the next step; judging whether the maximum value in the difference values of every two of T1, T2...Tn is smaller than N%*T1 or not, wherein N is larger than 1 and smaller than 200, and if yes, performing the next step; judging whether the maximum value in the difference values of every two of t1, t2...tn-1 is smaller than N%*t1 or not, and if yes, determining that the n sound samples are snoring sounds. The invention further relates to an anti-snoring device. Through the adoption of the method and the device, the snoring sound can be effectively determined, environmental noise can be effectively filtered, and accordingly, the anti-snoring purpose can be realized effectively.