Two experiments were conducted to compare the performance and behaviour of lactating sows and piglets in farrowing pens with crates or multisuckling systems (group housing of sows and piglets in the second half of lactation). All sows were farrowed in commercial accommodation based on farrowing pens with crates. In experiment 1, fifteen purebred Landrace or Large White sows and litters were recorded in crates as a control (C). Three replicates of six sows and litters were recorded in a multisuckling system (M) in which the sows and litters were group-housed from 2 weeks after farrowing until weaning. Experiment 2 involved 30 Manor Meishan sows (0·25 Meishan genes). The sows and litters were arranged into three treatments and two replicates with five sows and litters in each group. Treatments comprised a control farrowing pen with crate system and two multisuckling systems differing in degree of accessibility of the piglet creep area. Multisuckling 1 (Ml) had a solid creep front with an open doorway (0·5 X 0·9 m) for piglet access. In multisuckling 2 (Ml), the lower 30 cm of the solid creep front was removed to facilitate piglet access and allow visual contact with the sows. Piglets were weaned at 31 days in experiment 1 and 29 days in experiment 2. After weaning, piglets were moved to controiled-environment, fully slatted accommodation and monitored for 7 days in experiment 1 and for 12 days in experiment 2. Mortality rate in the two systems was not significantly different. However, some piglets in the M system were crushed after grouping (0·3 and 0·2 piglets per litter in experiments 1 and 2, respectively). Growth rate was reduced in M piglets in the week after grouping (256, 184 (s.e.d. 21·4) g/day, P < 0·01 for C and M respectively in experiment 1 and 243, 150 and 209 (s.e.d. 12·2) g/day, P < 0·01 for C, Ml and Ml respectively in experiment 2) but was higher after weaning (271, 313 (s.e.d. 35·7) g/day, P > 0·05 for C and M respectively in experiment 1 and 148, 280 and 222 (s.e.d. 15·0) g/day, P < 0·01 for C, Ml and Ml respectively in experiment 2). In consequence, piglet live weight at 1 week after weaning did not differ between treatments (9·33, 9·74 (s.e.d. 0·36) kg for C and M respectively in experiment 1 and 9·30, 914 and 9·53 (s.e.d. Oil) kg for C, Ml and Ml respectively in experiment 2). Sucking behaviour of M piglets in both experiments was severely disrupted (P < 0·01) on the day of grouping and the day after grouping. Despite synchronized suckling, M litters had a high incidence of cross suckling (> 50%) throughout lactation. M piglets spent more time than C outside the creep area, even with a more open creep (Ml). The Meishan sows tended to be more docile and fought less at grouping than the white breeds (0·39 and 1-71 fights per sow per h). Immediately after weaning, C piglets spent more time fighting (8·0 and 1·0 (s.e.d. 0·99) % of time, P < 0·01 for C and M respectively in experiment 1 and 4·51, 0·09 and 0·09 (s.e.d. 019) % of time, P < 0·01 for C, Ml and M2 respectively in experiment 2). These studies demonstrate that, in a multisuckling system, piglets achieved similar overall growth rate to C piglets, since both received checks in growth at different times and for different reasons.