The influence of reproductive variables on cervical cancer incidence, controlling for other sociodemographic factors, was estimated in Norwegian register and census data, using Poisson regression models. Among the 1.3 million women under observation, a total of 2,870 cases of cervical cancer were diagnosed. According to models restricted to parous women, parity level had no independent impact on cervical cancer incidence, but a clear effect of age at first birth was noted. It was most pronounced in the squamous cell carcinomas, where the incidence was reduced by 48 percent from age at first birth < 21 years to age at first birth 27+ years. Women without children had the same cervical cancer incidence as parous women with a first birth after age 24. The sociodemographic variables controlled for exerted a strong net effect on the cervical cancer incidence. Educational level was related inversely to the cancer risk. Moreover, an increased risk was seen for women who had given birth when they were still single (never married) and for those who were divorced/separated at the time of the last previous census. A fairly small excess risk was found to be associated with living in non-rural compared with rural areas.