This study extends the long-term temperature model proposed by Alaton et al. (2002) by taking into account ARCH/GARCH effects to reflect the clustering of volatility in temperature. The fixed variance model and the ARCH model are estimated using Taiwan weather data from 1974 through 2003. The results show that for HDD/CDD the call price is higher under ARCH-effects variance than under fixed variance, while the put price is lower. Although different pricing methods are employed in pricing weather options, the effects of mean and standard deviation on option prices are mathematically proved to be the same as those in pricing traditional financial derivatives using the Black-Scholes formula.