The aim of this work was to study the effect of a part-time collective housing of females raised together since their birth on their performance, use of space and behaviour. We used 40 females born on the same day (D0) and being suckled by 8 females until D35 (8 litters of 5 'sisters of milk'). The 'sisters of milk' were housed together, by litter, in the same housing until D84 and were followed during two reproductive cycles (from D84 to D245). At D84, 8 young females (1 per litter) were separated to be individually housed until D245 (I group; n=8 females). The others 32 females were housed in 8 modules of 4 individual housing that could be linked together via connection hatches between two housing (G group), inducing a large and partitioned habitat (36 388 cm² against 9 097 cm²). In G group, females were isolated from 4 days before to 17 days after the birth (D171 to D191 and D213 à D233 in 1 st and 2 nd cycle) and were grouped the rest of the time. Space use and social interactions were measured by direct observation twice a day and two days a week from D84 to D245 and by 40-min video recordings at D120 and D168. Behaviour along the day was measured at D151 by 26 direct observations throughout the day. Housing system had no effect on live weight nor fertility, but female mortality was higher in the G compared to I group (34 vs 0%; P<0.05). Group housing allowed the observation of positive social interactions (11% of total behavioural observations) but we also observed injuries (68% of females of G females throughout the experiment of which 19% had middle or severe injury scores). The number of positive interactions was high and the number of injuries low at young ages (from D84 to D170; P<0.05). On the opposite, the number of injuries in grouped females was higher during reproductive life than before first kindling (27 and 19% of severe wounds in 1 st and 2 nd cycle vs 2% before D170; P>0.05). Although in both groups, females were observed mainly on the floor (70% of observations), the number of vertical movements of grouped females was 3 times higher than isolated ones (1.8 and 1.2 vs 0.6 and 0.4 no./h in G vs I group at D120 and D168; P<0.05). These results suggested that the group housing of young females that knowns each other in a large partitioned space before their first kindling is a promising way to improve animal welfare. On the opposite, collective housing in lactating does is detrimental to animal health and should be avoided, even among females raised together since birth.