In stall-feeding conditions, food diversity has been shown to increase an animal's motivation to eat and its level of food intake. This effect has also been suggested, but not demonstrated, for free-grazing animals. In free-grazing conditions, however, animals may have to search for the preferred plant species and this may reduce their intake rate. Moreover, the management of diverse swards faces a risk of overgrazing the preferred species and undergrazing the less preferred species, leading to resource degradation. We investigated the effect of food diversity on foraging behaviour and intake in grazing sheep, in interaction with the type of diversity (conterminal monocultures v. mixtures) and management (two monocultures offered simultaneously v. successively over time). Five treatments were compared with five groups of five dry INRA 401 ewes for five periods, using a Latin-square design: LOL ¼ grazing a monoculture of Lolium perenne cv. Herbie (L); FES ¼ grazing a monoculture of Festuca arundinacea cv. Florine (F); STRIPS ¼ grazing conterminal monocultures of L and F, 0·5:0·5 by ground area, with animals having a free choice between both species; TIME ¼ grazing L from 16:00 to 09:00 h and F from 09:00 to 16:00 h; MIX ¼ grazing an intermeshed mixture of L and F. Sward height was maintained at 9 cm. Each period comprised a 5-day adaptation subperiod followed by a 5-day measurement subperiod. Dietary choices and intake were measured using n-alkanes, and grazing time was assessed using the Ethosys w recording system. Intake rate was calculated as the ratio of intake to grazing time. Intake in the monoculture treatments were higher than predictions made for housed animals given similar forages ad libitum, demonstrating that sward conditions were favourable for the expression of voluntary intake in all treatments. Intake was higher in LOL than in FES, this result being consistent with predictions for stall-feeding conditions. There was no difference in diet composition, foraging behaviour and intake between the bispecific treatments (STRIPS, TIME and MIX). Intake was higher in the bispecific treatments than in LOL (1666 v. 1492 g digestible organic matter per day, P , 0·001). Grazing time was higher in STRIPS and MIX than in LOL (420 and 430 min respectively v. 366 min, P , 0·01). Dry ewes increase their intake when offered a choice of herbage species at pasture, this being mediated via an increase in grazing time rather than an increase in intake rate. There was no interaction with type of diversity or type of management.