The invention discloses a carrier phase correcting method based on QAM modulation, which comprises the steps of: forming a signal set S1 after taking the biquadrate of all signals S0j in a received signal set S0, calculating a correcting angle of theta1-phi of the received signal corresponding to each point on the class1 in the S1, correcting the S1 signal so as to acquire an S2, calculating to obtain the minimum value of a real field of the received signal which is subjected to biquadrate extraction and corresponds to each point on the class 1, setting the minimum value as a selected threshold value limit, dividing the signals in the S2 into I, II and III three areas on the basis of the condition as follows: Rcostheta is less than limit, Rcostheta is more than limit and Rsintheta is morethan 0, Rcostheta is more than limit and Rsintheta is less than 0, performing the maximum likelihood fit on each received signal in the I, II and III areas according to three angles of phi, 4*atan(1/3) and -4*atan(1/3), calculating a second correcting angle beta, and correcting the angles of (theta1-phi+beta) of all the received signals in the signal set S0. In the method provided by the invention, the biquadrate operation and the correcting operation need not be performed twice, so the operand is greatly reduced, the dependence on hardware is reduced and the digital circuit linear-speed treatment is convenient to perform.