Safety stock calculations are difficult for products with intermittent demand, long production lead times, and high monetary values. Theoretically, forecasts can be used to reduce the need for safety stocks. A high precision forecast minimizes the need for safety stock and forecast evaluation measurements can be used to calculate the safety stock level. However, a more realistic determination of safety stock levels can he obtained by simulation. In this paper, simulation is used to model and experiment on a case with three end products in order to determine the relationship between safety stock levels and service levels. Also, a comparison is made with theoretically calculated safety stocks to see how well basic theoretical models for safety stock calculations fulfill the requirements of service level. The result is that simulation can provide a much more accurate determination of safety stock levels for intermittent demands than theoretical calculations.