The stochastic inventory routing problem involves the distribution of a commodity such as heating oil over a long period of time to a large set of customers. The customers maintain a local inventory of the commodity which they consume at a daily rate. Their consumption varies daily and seasonally and their exact demand is known only upon the arrival of the delivery vehicle. This paper presentes a detailed analysis of this problem incorporating the stochastic nature of customers' consumptions and the possibility of route failures when the actual demand on a route exceeds the capacity of a vehicle. A number of solution procedures are compared on a large set of real life data for a period of 12 consecutive weeks. The winning strategy, though computationally more expensive, provides the best system performance and reduces (almost eliminates) the stockout phenomena.