Pressures from consumers concerned with sow well-being in gestation stalls have encouraged producers to consider how to transition from individual gestation stalls to pens in existing buildings. The aim of this study was to evaluate the performance and well-being of sows in pens that were retrofitted from stalls (2.1 by 0.55 m). Large pens (5.5 by 7.3 m) housed 26 sows and small pens (5.5 by 1.7 m) housed 6 sows, with equal floor space allowance (1.5 m(2)/sow) in both pens. Floor space allowance in pens was set to ensure that pens and stalls occupied the same building space per sow. Data were collected from 815 sows (parity 1 to 8) in 13 large pens, 26 small pens, and 326 stalls that served as control. All sows were mated in stalls and moved to pens or stalls after pregnancy confirmation at 5 wk after mating. Daily feed allowance for sows was dropped on a solid floor in pens and was equal in all 3 housing systems. Sows stayed in their assigned gestation housing treatment for 1 pregnancy. Data were analyzed using the Proc FREQ and the Proc Mixed procedures of SAS. Sows in large pens gained less weight during gestation than sows in small pens and stalls (33.4 vs. 39.5 and 41.5 kg, SE = 1.64; P = 0.01). In addition, the proportion of sows assigned to the study that farrowed was the lowest in large pens (92%, χ(2) = 9.52; P < 0.01) and the highest in stalls (98%), with sows in small pens being intermediate (95%). Consequently, the highest percent of sows were removed from large pens (15.8%, χ(2) = 6.75; P < 0.05) and lowest percent of sows were removed from stalls (9.2%), with sows in small pens being intermediate (11.7%) during the study period. Causes for sow removal were different among the housing treatments. Poor reproductive performance was the top reason for culling sows in large pens and in stalls. In small pens, mortality was the top reason for sow removal. Housing treatment did not affect live litter size at birth (12.5, 12.2 and 12.3 piglets, SE = 0.21) or weaning (10.2, 10.1 and 10.3 piglets, SE = 0.20) for large pens, small pens, or stalls, respectively. Results suggest that the performance and well-being of sows were compromised in pens, as indicated by decreased farrowing rates and increased sow removal rates. The limited floor space allowance and the competitive floor feeding system could be major contributors to the compromised performance and well-being of group-housed sows in this study.