The acquisition and processing of events from sensors or enterprise applications in real-time represent an essential part of many application domains such as the Internet of Things (IoT), offering benefits to predict the future condition of equipment to prevent the occurrence of failures. Many organisations already use some form of predictive maintenance to monitor performance or keep track of emerging business situations. However, the optimal design of applications to allow an effective Predictive Mainte-nance System (PMS) capable of analysing and processing large amounts of data is only scarcely exam-ined by Information Systems (IS) research. Due to the number, frequency, and the need for near-real-time evaluation systems must be capable of detecting complex event patterns based on spatial, temporal, or causal relationships on data streams (i.e. via Complex Event Processing). At the same time, however, due to the technical complexity, available systems today are static, since the creation and adaptation of recognisable situations results in slow development cycles. In addition, technical feasibility is only one prerequisite for predictive maintenance. Users must be capable of processing this vast amount of data presented without considerable cognitive effort. Precisely this challenge is even more daunting as op-erational maintenance personnel have to manage business-critical decisions with increasing frequency and short time. Research in Human Factors (HF) suggests Situation Awareness (SA) as a crucial sys-tem’s design paradigm allowing human beings to understand and anticipate the information available effectively. Building on this concept, this paper proposes a PMS for promoting operational decision makers’ Situation Awareness by three design principles (DP): Sensing, Acting, and Tracking. Based on these DPs, we implemented a PMS prototype for a scenario in Oil and Gas pipeline operations. Our finding suggest that the use of SA is of particular interest in realizing effective PMS.