The purposes of this study were to evaluate subterranean clover-based leys on farms and in experiments using several pasture parameters, and to assess the impact of winter cleaning on the productivity and botanical composition of clover swards. Annual pastures were monitored on a group of 5 farms in the Wagga district and compared with an experimental subterranean clover (Trifolium subterraneum L.) pasture. The major problem in the farm paddocks was a lack of legume biomass due to poor legume densities, a consequence of the use of the soft-seeded cultivar Woogenellup and a high content of grassy weeds. The farmers in the group were unaware of the tools, parameters and benchmarks for making quantitative pasture comparisons. In 2 experiments, a range of subterranean clover swards were generated through the use of cultivars, seeding rate and winter cleaning treatments, grazed at 15 sheep/ha and monitored for 3 years. Appropriate benchmark values for the seed pool of subterranean clover were 300'350 kg/ha in winter and 600'700 kg/ha in summer. On the basis of both winter production, a function of May seedling density (target >1000 seedlings/m2) and spring production, which depended on the cultivar maturity, Junee was superior at Wagga to either Dalkeith (earlier maturing) or Woogenellup (softer seeded). Winter cleaning, using selective herbicides (fluazifop, simazine) to remove grasses and weeds, was advantageous in achieving a high content (>90%) and productivity of subterranean clover, provided that the legume content of the pasture was at least 28%, or >20% of total ground area before herbicide application in winter. In winter-cleaned swards, legume growth increased by up to 80%, legume biomass was improved by up to 46% and legume content increased from <50 to >95%. The main disadvantages of winter cleaning were increased areas of bare ground and reduced total biomass for several weeks after herbicide application, and the rapid development of ryegrassthat was resistant to at least 1 of the herbicides used. The strategic use of observations to monitor the performance of pastures and their response to management is discussed.