This paper focuses on the forecasting risk analysis in supply chains with intermittent demand, which is typical for the inventory management of the 'slow-moving items', such as service parts or high-priced capital goods. The adopted demand model is based on the Generalised Beta-Binomial Distribution (GBBD), which is capable of incorporating the additive distortions in the demand historical records as parameters. For this setting, there are proposed explicit expressions for forecasting risk and the prediction function, which minimises the error impact on the risk. The efficiency of the proposed approach is confirmed by computer simulation and is illustrated by an application example for forecasting of the intermittent demand values for car spare parts.