This work provides empirical examination of conditionality, degree of persistence of international coffee price volatility and effects of policies in Tanzania. In modelling coffee price volatility, previous studies treated variation in price as unconditional and volatility addressed as constant overtime but recent studies indicated that price volatility exhibits auto-regressive conditional heteroskedastic behavior. Noting that, this study has attempted to model coffee price volatility using the GARCH (1,1) process. It was written out of the time-series data on the international coffee price paid to growers in Tanzania spanning from 1980 through 2008. The study establishes that coffee price had no regular pattern rather a random walk process, non-conditional and its persistence overtime displays memory decay. Changes in trade policies within Tanzania do not have any effect on the world coffee price. The likely explanation is that Tanzania is a price taker, cannot affect settings of international prices. The policy response to this include new marketing approach, ethical trading, Tanzania coffee can be marketed as a brand and revitalization of cooperatives.