Through comparison of the evaluation of climate prediction operation for Gansu，Qinghai，Shaanxi and Ningxia in 2009 by the
old and new evaluation methods respectively，for precipitation and temperature，it has been found that both the average annual score
and the maximum score by the old method were systematic higher about 15% than that by the new method，but the minimum score by
the old method was systematic lower than that by the new method in most cases． It was relatively higher probability to reach full mark
for the maximum sore by the old method especially to temperature prediction，and less probability by the new method． A zero score was
high probability by the old method when the prediction was opposite to the fact，but 20 points by the new method at least． The new evaluation
method for climate prediction was more rigorous than not only the old method，but also the evaluation method for weather forecast．