As part of a national evaluation series, five grass species were oversown with white clover (Trifolium repens L.) into glyphosate-treated plots on a summer-dry northern face of the Hokonui hills, Southland. Swards of these species were compared with the resident pasture, under an annual fertiliser input of 0 or 250 kg superphosphate (0-9-0-11)/ha per year and several 7- or 30-day grazing periods/year. Nui ryegrass (Lolium perenne L.) had the highest annual and seasonal pasture yields, especially in spring when pasture yields were at least 18% better than any other grass species. Wana cocksfoot (Dactylis glomerata L.) persisted on all slopes and had similar pasture yields to Nui in summer in the first 3 years and in autumn in the final 3 years. Maru phalaris (Phalaris aquatica L.) established slowly but spread and persisted to equal the yield of Wana in Year 5 and Nui in Year 6. Although the yield of Matua prairie grass (Bromus willdenowii Kunth) was second only to Nui in the first 3 years, it declined rapidly after the management was changed from 2-day grazing to 7- or 30-day grazing. Roa tall fescue (Festuca arundinacea Schreb.) did not establish well and never contributed more than 12% to the average pasture yield. The resident pasture always had the lowest annual and seasonal yields, being up to 30% less than Nui pastures. The fertiliser and grazing management treatments had no effect on the relative rankings of the grasses used. Withholding fertiliser resulted in a significant drop in production of 15-20% for the third and subsequent years. Both the sown species and white clover contents were reduced by withholding fertiliser. The 30-day grazing periods significantly reduced pasture productivity by up to 23% when compared to 7-day grazing periods, especially in winter and spring.