The objective was to compare gestation length in chronically instrumented (laboratory) pregnant sheep (n = 131) and in the breeding flock (n = 476) that provided the experimental sheep. In the breeding flock, gestation length was normally distributed and varied between 141 and 151 days (mean = 147 ± 0.1 days). In the laboratory sheep, gestation length varied between 128 and 151 days (mean = 142 ± 1 day), and was bimodal, with 35.9% delivering preterm (<141 days). To examine potential factors that contributed to the preterm birth, a severity score was used, which comprised surgery characteristics, number of experiments and maternal or fetal complications. There was a significant inverse linear relationship (P < 0.001) between the total severity score and gestation length. The median values for the surgical (15 v. 12), overall complication (6 v. 2), maternal complication (2 v. 0) and fetal complication (2 v. 2) components were significantly greater in the preterm compared with the term groups. There was no relationship between fetal number and gestation length in either group. It is concluded that in chronic pregnant sheep preparations, there is a significant incidence of preterm birth and that this is associated with the severity of the surgical intervention and with several maternal and fetal complications.