OBJECTIVE--To investigate the incidence of spontaneous abortion in a population of women in order to establish their risk of spontaneous abortion and the obstetric factors predisposing to it. DESIGN--Prospective study of women recruited by radio and poster appeal and from hospital outpatient clinics. SETTING--English provincial community. PATIENTS--630 Women from the general population intending to become pregnant. INTERVENTIONS--The viability of the pregnancy was assessed by abdominal ultrasonography before completion of the eighth week, and the assessment was repeated if vaginal bleeding occurred. MAIN OUTCOME MEASURE--Spontaneous abortion or live births in women with or without a previous history of spontaneous abortion. RESULTS--The overall incidence of clinically recognisable spontaneous abortion before 20 weeks of gestation was 12% (50/407 pregnancies). The risk of spontaneous abortion in each category of patient was classified with respect to the patient's past reproductive performance and found to be influenced greatly by her previous obstetric history. In primigravidas and women with a history of consistently successful pregnancies the incidences of abortion were low (5% (4/87) and 4% (3/73) respectively), whereas women with only unsuccessful histories had a much greater risk of aborting the study pregnancy (24% (24/98)), even when their sole pregnancy had ended in abortion (20% (12/59)). The outcome of the last pregnancy also influenced the outcome of the study pregnancy; only 5% of women (5/95) whose previous pregnancy had been successful aborted, whereas the incidence of loss of pregnancy among women whose last pregnancy had aborted was 19% (40/214). CONCLUSIONS--A knowledge of the patient's reproductive history is essential for the clinical assessment of her risk of spontaneous abortion. As the most important predictive factor for spontaneous abortion is a previous abortion, the outcome of a woman's first pregnancy has profound consequences for all subsequent pregnancies.