The dependability of any kind of context-aware pervasive system inherently relies on the ability to detect and recognize context events robustly. However, due to the inaccuracy of real-world sensors the accurate recognition and handling of context information is a fundamental problem leading to ambiguities and inconsistent behavior of applications. Adaptable Pervasive Flows (APF) are a novel workflow-based programming paradigm for pervasive applications. An APF encodes the temporal dependencies of a user’s tasks. We propose a Fuzzy Event Assignment (FEvA) algorithm that exploits this flow-knowledge to significantly improve the robustness of context-aware pervasive applications by resolving the ambiguities inherent to the context data. Our experiments show that FEvA reaches an event assignment accuracy of 78% to 97% and improves the performance of dealing with false positive, out-of-order events, and missed context information.