We attempted to determine if an amended social security system has lessened the rate of sickness absence during pregnancy in Sweden over the period 1978-1989. We studied the records of 3998 women who gave birth in 1978, 1986, 1988, and 1989 at the University Hospital in Linköping and the Värnamo County Hospital. Retrospective collection of all data concerning the rates and durations of sickness absence during pregnancy, drawing of parental benefit, and use of granted pregnancy benefit was performed from Sweden's standardized social security files. Obstetric variables concerning the course and outcome of pregnancy and delivery, as well as the health status of the newborns, were obtained from standardized and antenatal care and delivery files. Between 1978 and 1989, the rate of sickness absenteeism during pregnancy increased by almost 100% for periods of absence not supported by a doctor's certificate and by about 50% for those with a doctor's certificate. During the same time, the average number of days of sick leave per pregnant woman more than doubled. The changes were most apparent among younger pregnant women. During the decade studied, no significant differences were found with regard to antenatal care, modes of delivery, or the health status of the newborns. In spite of the introduction within the Swedish social security system of more generous rules for pregnant women, the changes in the rates of registered sick leave during pregnancy observed over time were most unfavorable. The present study indicates that the rate of sick leave during pregnancy reflects a complex social phenomenon and cannot be explained solely by an increase in the rate of actual illness or sickness. Such a rate is also likely to depend on general attitudes and expectations among pregnant women, which may vary over time. When amendments in social benefits for pregnant women are considered to improve the health status of the pregnant population, society ought to clearly define the precise aim pursued and the consequences expected to escape unintended socioeconomic results.