We studied the effect of vaginal progesterone (P) treatment during the luteal phase of patients who had had a tubal pregnancy (TP) and were planning another, in a prospective, randomized, double-blind trial. The outpatient clinics of two University hospitals and three central hospitals had 135 patients treated for tubal pregnancy: 100 with grossly normal fallopian tubes (supposing an accidentally abnormal luteal phase as a possible etiology of their first TP) and 35 with signs of earlier pelvic inflammatory disease (PID etiology). They were treated with vaginal P (25 mg b.i.d.) or placebo during cycle days 16-24, for 10 months. Serum P levels after a single vaginal or oral dose were compared. The rates of conception, delivery, spontaneous abortion and recurrent TP were recorded, and fetal and placental weight measured. Both vaginal and oral formulas of P provoked a physiological (24-43 nmol/l) rise in serum concentrations. P and placebo-treated cycles resulted in a nearly equal number of pregnancies (33/37 resp.). Of the 55 infants born 53 were to mothers without signs of earlier PID (53/100); only 2 (2/35) to mothers in whom signs had been present. Recurrent TP occurred in 9% of all pregnancies. Four out of six recurrent TPs were patients with signs of PID (4/35), but two were without such signs (2/100): one occurred during placebo and one during P-treated cycle. Prophylactic P treatment of patients at risk of recurrent TP does not improve fertility or prevent recurrent TP. This indicates, that the functional etiology of recurrent TP, as compared to infection, is not important.