Grazing management has been identified as a means of increasing livestock production and improving the composition of perennial pastures. The benefits of intensive rotational grazing have been the subject of much debate, but few studies have evaluated contrasting management of intensive rotational systems. A grazing management experiment was established on a pasture with cocksfoot (Dactylis glomerata L., varieties Porto and Kara) as the dominant species, to investigate different stocking rates, paddock numbers and rotation speeds, and a flexible treatment that adjusted grazing time, rest periods and stock numbers for optimal pasture utilisation. Data were collected on pasture composition and diet quality assessed by using faecal analysis, animal weight changes and pasture characteristics. Animal production per hectare was greatest for fast rotations (56 days'rest) at high stocking rates (HStR, 13.6 dry sheep equivalents (DSE) ha–1 ), but continuous grazing (CG) was equally productive. Although flexible grazing based on the 3–4-leaf stage was proposed as the best balance between pasture production and quality, this treatment had lower stocking rates (9.2 DSE ha–1 ) and was not as productive. No treatment negatively affected pasture composition over the 4-year period. Area of bare ground was highest for the HStR CG treatment" however, the 30-paddock rotations were able to limit bare ground at the same stocking rate. The results indicated that intensive rotational grazing could be effectively managed by using green herbage allowance. In spring, green herbage allowance needed to be 1–1.5 kg green dry matter (DM) DSE–1 day–1 , which increased to 5 kg green DM DSE–1 day–1 as the quality of green DM decreased, to allow selective grazing to enhance diet quality.