OCCUPATIONAL APPLICATIONS Designing sustainable cyclic work requires attention to both the workload amplitude as well as the duty cycle, the fraction of the work cycle with active workload, that therefore also defines the recovery phase of the cycle. A number of different approaches and models have been developed to calculate the required recovery time for a given load and duty cycle. We present a comparison of three types of models at the "breakpoint" that defines the boundary of load amplitude and duty cycle where fatigue begins to accumulate faster than recovery allows within the work cycle. This comparison shows considerable variation between models of the "allowable" load or duty cycle depending on the method used. Practitioners should thus be cautious applying these models indiscriminately in job design as their results can vary substantially. In particular, differences between the tasks used for model formulation and application may compromise validity, and model application in a given context should be verified before broad application. TECHNICAL ABSTRACT Rationale: There is a need for tools to help design sustainable work in which muscular capacity and other human resources can recover at least as quickly as they are used. Purpose: In this brief report, three different approaches presented in the literature to determining work-rest schedules in cyclic work are compared. Methods: First, a set of five different muscular endurance models coupled with a recovery time model were considered, both with and without a dynamic work correction factor. Second, we examined a model of "resumption time", and third a psychophysically-based model of maximum duty cycle was included. These models were compared using the concept of a "breakpoint" in fatigue accumulation-the point at which a given load amplitude and duty cycle combination begins to cause accumulation of fatigue in each cycle and from which there is inadequate time to recover. Results: While the five endurance time models all behaved similarly, both with and without the static-to-dynamic correction factor applied, the three different types of modeling approaches provided substantially different response patterns. The psychophysically based model provided the most protective guideline among the models compared. Conclusion: These models should be applied with caution to particular work scenarios. Further research is needed to test accuracy and effectiveness when applying such models to a range of task scenarios to establish safe workloads and loading times in the design of repetitive work.