Grazing-system experiments address complex interactions among animals, pastures, soils, climate and management. As part of the national EverGraze program, a grazing-system experiment was designed to determine how the intensity of grazing management, from continuous grazing (P01) to flexible 4- and 20-paddock rotational systems (P04 and P20), influences the profitability and sustainability of a Merino ewe, terminal sire, lamb production system grazed on heterogeneous native pastures. When implementing such an experiment, it is important to understand and characterise landscape variability, and include this in the design of the experiment. A second challenge for grazing-system research is to operate experimental systems with sufficient flexibility to adequately represent commercial production systems and maintain even utilisation across treatments. The present paper addresses the following two issues: (1) the process used to characterise the potential productivity of variable native pastures and the results of this characterisation; and (2) the development of flexible systems that adequately represent commercial production within an experiment. This was undertaken with input from a project-steering committee called the EverGraze Regional Group, comprising producers, extension staff and private consultants. Prior to the commencement, the site was mapped into three production zones, namely, high (HPZ), medium (MPZ) and low (LPZ), by visually estimating green herbage mass in late spring and marking boundaries between zones with a GPS. The production zones represented differences in soil properties (gravel, pH and available P) and pasture composition, and were used to balance potential production among plots within the same replication. Grazing-system options were evaluated using the sustainable grazing systems pasture model to help choose an appropriate starting stocking rate. The initial stocking rate chosen for the spring-lambing systems was 5.4 ewes/ha. The modelling predicted large variations in feed availability and quality over summer among years; flexible management criteria were therefore developed, including variable sale time for lambs, to utilise the greater feed supply in better seasons. Minimum-pasture benchmarks (>0.8 t DM/ha standing herbage mass and >80% ground cover) and variable green herbage-mass targets were designed to sustain high levels of livestock production and prevent pasture degradation. Criteria for adjusting ewe numbers were developed, but were constrained to pre-joining (March), scanning (July) and post-weaning (December), being consistent with commercial practices. The experiment incorporated flexible management rules as these were considered integral to the successful management of commercial grazing systems.