Abstract Overall Equipment Effectiveness (OEE), which is commonly used for production performance monitoring, can mislead decision makers with low accuracy when a large amount of losses remains unclassified. This paper is to study the critical factors and potential pitfalls when trying to automatically estimate the OEE of a manufacturing system, considering uncertainty. Two methods based on fuzzy arithmetic and interval arithmetic respectively are proposed to manage the uncertainty in estimating the production speed, the stoppage duration, and the quality losses. Datasets from real-world settings are used to illustrate the concepts and the benefits of the methods in practice.