The survival of perennial ryegrass (Lolium perenne L.) plants was studied in sheep pastures in south-western Victoria during the dry summer of 1996–97. Recruitment of perennial ryegrass seedlings into the pasture sward was also monitored in the autumn–winter periods in 1997 and 1998. The objective was to investigate whether a tactical stocking strategy, consisting of variable length summer, autumn and winter rotations and continuous stocking in spring, might increase perennial ryegrass tiller survival and seedling recruitment in the autumn, compared with continuous stocking all year. The grazing strategies were compared on 2 contrasting pastures: an upgraded pasture [sown with newer cultivars of perennial ryegrass and subterranean clover (Trifolium subterraneum L.) with 26 kg phosphorus/ha.year as applied fertiliser] and a naturalised perennial ryegrass pasture receiving 6 kg P/ha.year. Paddocks were grazed by Border Leicester × Merino ewes, mated to a terminal sire so as to lamb in September. Perennial ryegrass tiller density was higher on the upgraded pasture with a mean density of 7750 tillers/m2 in early summer which declined to zero live tillers by mid summer. Live tillers began to reappear before the opening rains and then increased after the rain. Mean tiller density in the upgraded pasture declined over the 2 summers, with only 2050 tillers/m2 being present 2 months after the opening rains in 1998. There were no effects (P>0.05) of pasture type or grazing strategy on the number of tagged tillers that survived the summer period. Only 12% of the vegetative tillers, randomly tagged in December 1996, survived to May 1997. More than half of the tillers (56%) that produced a seedhead produced daughter tillers which survived the dry summer–autumn period. A significant (P<0.05) interaction between grazing strategy and pasture type occurred with the number of perennial ryegrass seedlings that had established 4 weeks after the opening rains in 1997. There was a 5–11-fold increase in seedling numbers which regenerated in the tactically stocked, upgraded pasture compared with the other treatments. Seedling recruitment was considerably lower in the autumn of 1998, due presumably to an overall decline in perennial ryegrass density relative to annual grasses in 1997. A second experiment investigated the effect of excluding sheep from grazing at anthesis until seedhead maturation or until the opening rains, together with a mechanical seed dislodgment treatment at seed maturity. All exclusion treatments increased seedling recruitment 4–7-fold, compared with continuous stocking. The results suggest a possible mechanism by which perennial ryegrass density can be increased without expensive reseeding.