A breeding experiment in Coopworth sheep where selection was applied to backfat depth was evaluated after 14 years (1981-94). There were three self-replacing lines, two selected for increased or decreased subcutaneous fat depth, and an unselected control (average size = 51 ewes per line per year). Reproductive traits, ewe body weight and ewe fleece weight were monitored in each line as potential correlated responses. In the last 4 years studied, the fat (F), control (C) and lean (L) lines differed significantly in the proportion of ewes lambing (0·87, 0·93, 0·94 respectively), litter size at birth (1·64, 1·83 and 1·89), and lamb survival up to weaning, unadjusted for birth rank and sex (0·88, 0·81 and 0·74). The overall reproductive trait, lambs weaned per ewe joined (1·22, 1·33 and 1·26), did not differ significantly among lines. From pre-mating weights of ewes present in 1993 or 1994, means for mature ewes by line were 60·1, 62·4 and 63·3 kg, respectively (P < 0·01). Means for weight at scanning time (7 months of age) by line, as a proportion of mature weight were: 0·533, 0·561 and 0·559, respectively (P < 0·05). There were also significant line differences in ewe fleece weights, with the 1993 and 1994 data averaging 3·93, 4·05 and 4·27 kg respectively (P < 0·01). For weight of lamb weaned per ewe weaning at least one lamb, the L line had a 0·09 higher mean proportionally than the F line (P < 0·05), but for a composite trait which was a weighted combination of weight of lambs weaned and ewe fleece weight per ewe joined (with or without allowance for different ewe live weights), the L line was no longer significantly different from the F line. Repeatabilities and single-record heritabilities were also estimated for various ewe traits. The implications are that lean-lamb selection would be associated with only small changes in net reproduction, ewe fleece weight and live weight, but larger differences would be expected from selection in the opposite direction.