We have sought to clarify the definition and importance of newly diagnosed thrombocytopenia in pregnant women by establishing an appropriate reference interval for the platelet count, and by observing the outcome in a cohort of thrombocytopenic pregnant women. The mean platelet count was lower in 2,155 healthy pregnant than non-pregnant women, and varied with race, but did not fall with increasing gestation, except in Black women. 101 of approximately 2,800 women became newly thrombocytopenic during pregnancy by conventional criteria (platelets less than 150 x 10(9)/l), without an apparent cause. Using the reference intervals established during the study, however, 24 of these women would not have been regarded as thrombocytopenic. No bleeding tendency was observed in the mothers or their infants. Maternal platelet counts became normal in the postnatal period. Thrombocytopenia occurring for the first time during pregnancy may be a different condition from auto-immune thrombocytopenia, and we suggest the term 'pregnancy-associated thrombocytopenia' (PAT). PAT does not appear to be epidemic; it does not threaten the fetus, but it cannot be distinguished from auto-immune thrombocytopenia in individual cases.