The effect of attending breech, twin, and post-date pregnancies on home birth outcomes was assessed. The same form was used to collect data on a convenience sample of 4,361 home births attended by apprentice-trained midwives from 1970 to 1985 and 4,107 home births attended by family physicians from 1969 to 1981. Data sets were compared to find 1,000 pairs of pregnant women, one from each group, who were matched for age, sex, socioeconomic status, race, and medical risk. The perinatal mortality rate for the midwife-attended births was 14 per 1,000 (three fetal deaths before labor, six intrapartum fetal deaths, and five neonatal deaths). The perinatal mortality rate for births attended by family physicians was five per 1,000 (one fetal death before labor, two intrapartum fetal deaths, and two neonatal deaths). The difference was statistically significant; however, the differences disappeared when cases involving post-dates, twin, or breech deliveries were eliminated from the sample. Although the data are more than a decade old, they support the premise that outcomes for low-risk home births are comparably good whether attended by physicians or midwives. However, the findings do raise questions about the safety of attending high-risk births at home.