Aiming at the problem that the measurement accuracy of light addressab le potentiometric sensor (LAPS) is sensitive to temperature,a kind of temperat ure compensation method for LAPS is put forward in our study based on support vector machine (SV M).According to the measured data of the pH values of phosphate buffer saline in various temp erature conditions, the temperature compensation model of LAPS is established.The SVM classifies t he data by mapping non-linear relationship between the output saturation photocurrent and the temp erature to the high-dimensional space using kernel function,and the nonlinear learning proble m is transformed to linear learning problem in the high-dimensional space.The kernel method cou ld efficiently approximate the nonlinear temperature features of LAPS and improve the optimiza tion of the example temperature features.The experimental results show that in the case o f (20±1) ℃, (20±2) ℃ and (20±5) ℃,the standard deviation of the pH value in LAPS system output after temperature compensation declines to one-third,one-sixth and one-fourth of th e original value,respectively.The standard deviation of the pH value is less than 0.05pH using the temperature compensation model for LAPS.The temperature compensation model for LAPS could effectively reduce the influence of temperature and significantly improve the mea sureme nt accuracy of LAPS.