An organization must manage its resource consumption and material flows in order to satisfy the demand of its products as efficiently as possible. Managing of the aforementioned requires a balance between the organizations resources (such as the capability of distribution and production) and the market demand. According to Gardner (1990), an estimation of future demand is a necessity for maintaining the balance. An instrument that is used frequently to estimate future demand is demand forecasting. The demand forecasting practice has been thoroughly studied and a plethora of academic contributions exist on the topic. However, a best practice demand forecasting method does not exist for every kind of product. The purpose of this paper is to identify which time series forecasting method that will result in the lowest error rate on fast moving consumer goods. The methods are based on sales data of 18 articles from the company Coca-Cola Enterprises Sverige AB which predominantly sells soft drinks. The majority of the theoretical framework is time series models presented by the authors Stig-Arne Mattsson, Patrik Jonsson and Steven Nahmias. The paper identifies Exponential smoothing with individual input variables as the forecasting method with the lowest error rate. The method gave the lowest possible error rate on over 55 percent of the articles. In addition, the combined error rate of the articles using Exponential smoothing with individual input variables gave the lowest overall error.