The agronomic performance of a range of perennial ryegrass cultivar-endophyte combinations was compared in 16 trials conducted at sites throughout New Zealand. Each trial was run for 3 years according to seed industry evaluation protocols, measuring variables including: dry matter (DM) yield (total annual and seasonal DM yield), ryegrass ground cover at the end of 3 years, susceptibility to plant pulling, and rust incidence. The change in DM yield over the 3-year term of each trial was also analysed. There were significant differences among cultivars in total annual DM yield, and in seasonal DM yield for each of the five periods of the year among which yield was split (winter, early spring, late spring, summer, and autumn). Mean annual yield declined between Year 1 and Year 3 in all regions except Taranaki, by between 0.8 t DM/ha (Canterbury) and 5.3 t DM/ha (Waikato). There were significant region, year, region Ă year, and cultivar Ă year interactions in yield change. Significant differences in ground cover score for perennial ryegrass at the end of 3 years were recorded among cultivars. However, these differences did not mirror on-farm observations of ryegrass persistence in the Waikato during the drought of 2007/08, suggesting that the standard trial protocols currently used do not adequately test persistence. It is recommended that industry cultivar testing needs to more accurately assess perennial ryegrass persistence, using new protocols including running trials on commercial farms, for more than 3 years, and using mixed swards. Genetic differences in persistence could be generated more quickly by choosing sites known to challenge perennial ryegrass growth and survival. Keywords: perennial ryegrass, cultivar evaluation, DM yield, persistence testing, plant pulling, rust