As systems become increasingly more complex, operator workload frequently is the major limitation to system performance. System designers have employed a variety of metrics attempting to assess workload in various settings. Confusion exists as to data interpretation when the metrics used provide different assessments; that is, they dissociate. This paper looks at two metrics - performance and subjective ratings - and suggests a theory-based interpretation as to why dissociation occurs. The implications of this dissociation to system design are then briefly summarized.