Volatility estimation is an integral component of risk management, option pricing, and portfolio allocation. REIT volatility is examined using a Bayesian GARCH model. This paper discusses shortfalls of maximum likelihood estimation, which are commonly used for estimating GARCH models, and elucidates the advantages of the Bayesian alternative. A portfolio allocation problem highlights the differences in decision making from these methods. Conditional variance estimation uncertainty is found to increase with volatility.