Abstract: A single period inventory model based on the conditional value-at-risk (CVaR) was developed for risk aversion decision-maker. Only considering demand interval， mean and variance information， the maximum entropy approach was used to estimate the demand distribution for both of the two demand uncertainties. The results showed that the decision-maker should adopt the uniform and exponential distribution as the potential demand distribution when only knowing the demand interval， and mean and variance information. On this basis， the CVaR-based inventory strategies and performances were deduced. The simulated results showed that the inventory strategy derived from the estimated distribution by maximum entropy will lead to a certain performance loss， however the loss ratio is very limited. It indicates that the ordering strategy based on the maximum entropy has good robustness.

Key words: inventory model, maximum entropy approach, risk aversion, conditional value-at-risk, robustness