The invention discloses a method for testing a Seebeck coefficient of a thermoelectric material. The method integrates characteristics of a steady state method and a dynamic method, and mainly comprises the following steps: I, according to a dynamic testing method, keeping the temperature of a low-temperature end of a sample generally unchanged, rapidly increasing the temperature of a high-temperature end, recording a temperature difference delta T and a potential difference delta U of two ends in real time, and performing linear fitting delta U and delta T into the relationship that delta U=alpha 1*delta T+delta u; II, according to the equation, when delta T=0, acquiring delta U deviation, or when delta U=0, acquiring delta T deviation, and determining the two values as system errors; III, when keeping the average temperature T0 of the sample the temperature difference of two ends unchanged within 5-10K, respectively testing a potential difference delta U0 and a temperature difference delta T0 by using the steady state testing method, and according to an equation that alpha reality=delta U0 / delta T0-delta T deviation=delta U0-delta U deviation / delta T0, introducing delta U deviation or delta T deviation, thereby obtaining alpha reality. By adopting the method, precise testing on the Seebeck coefficient is achieved, system errors introduced in the testing process are eliminated, testing condition requirements are reduced, and very great learning significances can be achieved for making and optimization of data acquisition and processing modes of self-developed, to be developed and conventional commercial equipment.