This paper presents a study examining interruptionsin the wild by portraying the handling of interruptionsin manufacturing from a distributed cognitionlens. By studying how interruptions occur and are handledin the daily activities of a work team at a large foundry forcasting heavy diesel engines, we highlight situations whenthe propagation, transformation, and representation ofinformation are not supported by prescribed work processesand propose recommendations for how this can beamended. The study was conducted by several visits to theaforementioned factory with cognitive ethnography as thebasis for the data collection. The focus was on identifyinginterruptions and analysing these through a distributedcognition framework as an initial step towards studyinginterruptions in a manufacturing environment. The keyfindings include the identification of three, previouslyundefined, types of interruptions and the conclusion thatinterruptions do indeed affect the distributed workload ofthe socio-technical system and thus the overall productionperformance at the casting line.