An actual power curve fitting based theoretical power determination method for a wind power plant comprises: reading wind power plant data; preprocessing the wind power plant data; calculating wind speed at the height of a wheel hub of a fan; filtering preprocessed fan power data to remove fault data, establishing a sample data set for actual power curve fitting, and dividing the sample data set; calculating theoretical power of the fan; adopting a correlation coefficient weighting method for correcting the theoretical power of the fan; obtaining theoretical power of the wind power plant; and evaluating theoretical power recovery precision through a correlation coefficient and a secondary center distance, and generating a data recovery quality factor. With the method, the theoretical power recovery precision is high and the tedious and complicated calculation is avoided; and the method is simple to operate, easy to realize and high in applicability.