An Auto-Regressive Conditional Heteroskedastic (ARCH) regression model is applied to analyze the changes in monthly hog and major pork (loin, shoulder, belly) price data in Taiwan over the period 1991-1998. The empirical results of testing for Unit-Root are used to identify the stationarity of changes in hog and major pork price, and select a suitable lag by means of the partial autocorrelation function. Based on the estimation of ARCH model and the Lagrange multiplier test, this study found that volatility clustering phenomenon of local hog and pork price exists because of foot-and-mouth epidemic outbreak during the sample Period.