We undertook a retrospective comparative study of all patients with an unstable slipped capital femoral epiphysis presenting to a single centre between 1998 and 2011. There were 45 patients (46 hips; mean age 12.6 years; 9 to 14); 16 hips underwent intracapsular cuneiform osteotomy and 30 underwent pinning in situ, with varying degrees of serendipitous reduction. No patient in the osteotomy group was lost to follow-up, which was undertaken at a mean of 28 months (11 to 48); four patients in the pinning in situ group were lost to follow-up, which occurred at a mean of 30 months (10 to 50). Avascular necrosis (AVN) occurred in four hips (25%) following osteotomy and in 11 (42%) following pinning in situ. AVN was not seen in five hips for which osteotomy was undertaken > 13 days after presentation. AVN occurred in four of ten (40%) hips undergoing emergency pinning in situ, compared with four of 15 (47%) undergoing non-emergency pinning. The rate of AVN was 67% (four of six) in those undergoing pinning on the second or third day after presentation. Pinning in situ following complete reduction led to AVN in four out of five cases (80%). In comparison, pinning in situ following incomplete reduction led to AVN in 7 of 21 cases (33%). The rate of development of AVN was significantly higher following pinning in situ with complete reduction than following intracapsular osteotomy (p = 0.048). Complete reduction was more frequent in those treated by emergency pinning and was strongly associated with AVN (p = 0.005). Non-emergency intracapsular osteotomy may have a protective effect on the epiphyseal vasculature and should be undertaken with a delay of at least two weeks. The place of emergency pinning in situ in these patients needs to be re-evaluated, possibly in favour of an emergency open procedure or delayed intracapsular osteotomy. Non-emergency pinning in situ should be undertaken after a delay of at least five days, with the greatest risk at two and three days after presentation. Intracapsular osteotomy should be undertaken after a delay of at least 14 days. In our experience, closed epiphyseal reduction is harmful. Cite this article: Bone Joint J 2015;97-B:412-19.