Fatigue has increasingly been viewed by society as a safety hazard. This has lead to increased regulation of fatigue by governments. The most common control process has been compliance with prescriptive hours of service (HOS) rule sets. Despite the frequent use of prescriptive rule sets, there is an emerging consensus that they are an ineffective hazard control, based on poor scientific defensibility and lack of operational flexibility. In exploring potential alternatives, we propose a shift from prescriptive HOS limitations toward a broader Safety management system (SMS) approach. Rather than limiting HOS, this approach provides multiple layers of defence, whereby fatigue-related incidents are the final layer of many in an error trajectory. This review presents a conceptual basis for managing the first two levels of an error trajectory for fatigue. The concept is based upon a prior sleep/wake model, which determines fatigue-risk thresholds by the amount of sleep individuals have acquired in the prior 24 and 48 h. In doing so, managing level 1 of the error trajectory involves the implementation of systems that determine probabilistic sleep opportunity, such as prescriptive HOS rules or fatigue modelling. Managing level 2, requires individuals to be responsible for monitoring their own prior sleep and wake to determine individual fitness for duty. Existing subjective, neurobehavioral and electrophysiological research is reviewed to make preliminary recommendations for sleep and wake thresholds. However, given the lack of task- and industry-specific data, any definitive conclusions will rely in post-implementation research to refine the thresholds.