To evaluate the prevalence of diabetes mellitus in pregnancy in Yup'ik Eskimos. A retrospective review of consecutive birth-log data and medical records was conducted. Records were taken from the primary care and tertiary referral facilities in Alaska Area Native Health Service. Some 630 consecutive deliveries were reviewed from 1 March 1987 to 29 February 1988, with 25 excluded for ancestry other than Alaska Native. Another 605 Alaska Native patient charts were reviewed, with 545 Alaska Native patients screened for diabetes in pregnancy. The study population had a mean age of 25.6 yr, mean gravidity of 3.4, mean parity of 1.9, and mean birth weight of 3567 +/- 493 g. Patients were screened with a 50-g glucose oral load with a plasma glucose 1 h later. 156 of 605 (25.7%) patients, with a screen > or = 7.8 mM received a 100-g OGTT. Of 605, 35 (5.8%) patients met O'Sullivan criteria, and 2 of 605 (0.3%) patients met WHO criteria for previous diabetes mellitus, for a total 37 of 605 (6.7% [corrected]) women with diabetes in pregnancy. The subjects who met O'Sullivan criteria had statistically greater mean age (29.9 yr), gravidity (4.9), parity (2.9), and birth weight of their infants (3678 +/- 389 g), compared with women with a screen < 7.8 mM. The prevalence of diabetes in pregnancy among Yup'ik Eskimos is twice the rate for the U.S. for all races, despite the Yup'ik having the lowest rate of diabetes mellitus among Alaska Natives. This may represent a large number of undiagnosed patients with impaired glucose tolerance, and may reflect the wide-spread dietary and life-style changes that have occurred in the Yup'ik in the last 30 yr. The Yup'ik present a unique opportunity to apply prevention techniques in a population with an emerging problem with glucose tolerance.