In order to understand better the factors which influence the persistence or non-persistence of Lolium perenne and Trifolium repens on damp hill land in a wet climate, swards in mid-Wales which had been converted from semi-natural rough grazing (predominantly Molinia caerulea or Nardus stricta or Agrostis/Festuca/Nardus) in 1961–81 were examined in 1982 and 1999, using a point quadrat technique. There was a marked decline between 1982 and 1999 in the proportion of L. perenne in the majority of the study areas. On the other hand, in 1999 the oldest swards (sown 30–38 years before) contained at least as high a proportion of L. perenne (22% of sward on average) as the younger swards (sown 18–29 years before). In two swards sown 25–26 years before, at 390–445 m above sea level, which received 95 kg N/ha per year, L. perenne was 36% of sward. On the other hand, no L. perenne was recorded on three swards which were no longer receiving fertilizer and which were being grazed at a greatly reduced stocking rate. The proportion of T. repens was rather low and declined between 1982 and 1999 (from 10 to 4% of sward on average). The application of more lime and more K and perhaps less N would probably have increased the proportion of T. repens. On at least one farm, the use of MCPA against Cirsium spp. had probably reduced the proportion of T. repens. There were positive correlations between the proportion of L. perenne and the proportions of T. repens and Poa spp. There were negative correlations between L. perenne and each of Agrostis spp., Festuca spp. and Holcus lanatus and also between T. repens and Agrostis spp. The proportion of Poa spp. was positively correlated with exchangeable Ca, nitrate-N and total N in soil and negatively correlated with exchangeable acidity.