In field trials at Viterbo in 1984-87, growth of Lolium perenne cv. Cropper, Dactylis glomerata cv. Dora, Festuca arundinacea cv. Maris Kasba and Phleum pratense cv. Kampe was monitored in Jan.-July each year using the Corrall and Fenlon cutting regime. Meteorological data were also collected. Growth rate was a linear function of temperature when water deficit did not occur; this relationship also held below a certain critical temperature when there was a slight water deficit. Maris Kasba could grow at lower temperatures but Cropper and Maris Kasba had a lower response to increased temperature than Dora or Mape. The data were used in conjunction with climatological data from Viterbo, Todi and Maccarese to construct a model enabling determination of the onset and duration of spring grazing. The model indicated that growth continued during winter in almost all years at the Mediterranean location (Maccarese) but that growth was always checked in winter at the inland site of Todi. The frequency of early cessation of growth in summer was greater at Maccarese than at the other sites. The grazing period ended in mid-July at all sites. The duration of spring grazing depended on when grazing ceased in summer at Maccarese but mainly depended on the earliness of the onset of spring growth at the inland sites.