When monitoring compliance of event streams at runtime, it is usually assumed that the events are received in the order in which they are produced. However, in reality, event data is typically transmitted from distributed sources and may be received out-of-order as a result. This is problematic for the compliance analysis in terms of accuracy, latency and computational efficiency. In this paper, we present an approach to manage these problems by exploiting knowledge of the event sources to identify the reliable prefix of the event stream that is guaranteed to be in order, allowing a reliable analysis on that prefix. Subsequently, we use the prefix alignment of the observed event stream and a process model to quantify its reliability and identify events that might arrive out-of-order, allowing to calculate a confidence score of compliance results. This improves efficiency and provides compliance results with awareness of potential uncertainty, enabling decision makers to respond appropriately to the identified issues.