It’s usually considered to be practical and effective to predict the amplitude of solar cycle using statistical precursor methods based on the geomagnetic precursors, but it will appear biggish error in a certain cycle prediction by this method. Particularly, the error exceeds 30% in predicting the amplitude of the 23rd solar cycle, which we are now standing. This paper provides a synthesis prediction method based on multi-information. Computer simulation shows that the new method is more adaptive and stable, compared with the statistical precursor methods based on the geomagnetic precursors. In predicting the 23rd solar cycle amplitude, the average error is only 10%.