BACKGROUND: Tubal ectopic pregnancy can be surgically treated by salpingectomy, in which the affected Fallopian tube is removed, or salpingotomy, in which the tube is preserved. Despite potentially increased risks of persistent trophoblast and repeat ectopic pregnancy, salpingotomy is often preferred over salpingectomy because the preservation of both tubes is assumed to offer favourable fertility prospects, although little evidence exists to support this assumption. We aimed to assess whether salpingotomy would improve rates of ongoing pregnancy by natural conception compared with salpingectomy. METHODS: In this open-label, multicentre, international, randomised controlled trial, women aged 18 years and older with a laparoscopically confirmed tubal pregnancy and a healthy contralateral tube were randomly assigned via a central internet-based randomisation program to receive salpingotomy or salpingectomy. The primary outcome was ongoing pregnancy by natural conception. Differences in cumulative ongoing pregnancy rates were expressed as a fecundity rate ratio with 95% CI, calculated by Cox proportional-hazards analysis with a time horizon of 36 months. Secondary outcomes were persistent trophoblast and repeat ectopic pregnancy (expressed as relative risks [RRs] with 95% CIs) and ongoing pregnancy after ovulation induction, intrauterine insemination, or IVF. The researchers who collected data for fertility outcomes were masked to the assigned intervention, but patients and the investigators who analysed the data were not. All endpoints were analysed by intention to treat. We also did a (non-prespecified) meta-analysis that included the findings from the present trial. This trial is registered, number ISRCTN37002267. FINDINGS: 446 women were randomly assigned between Sept 24, 2004, and Nov 29, 2011, with 215 allocated to salpingotomy and 231 to salpingectomy. Follow-up was discontinued on Feb 1, 2013. The cumulative ongoing pregnancy rate was 60·7% after salpingotomy and 56·2% after salpingectomy (fecundity rate ratio 1·06, 95% CI 0·81-1·38; log-rank p=0·678). Persistent trophoblast occurred more frequently in the salpingotomy group than in the salpingectomy group (14 [7%] vs 1 [