Studies carried out in foreign countries (US and UK, mainly) indicate that maternal characteristics, such as age, parity, social class, and prenatal care, are related to child's growth, mortality, and morbidity, as well as to cigarette smoking. These characteristics may act as confounding variables in the analysis of the effects of maternal smoking on babies in fetal and neonatal periods. Until now there has been a lack of information on the subject, because even the most recent available data concern women over age 14 regardless of obstetric history. This paper deals with smoking habits, before and during pregnancy, of 37,664 women included in a multicenter survey of perinatal preventive medicine (MPPI), which was performed in 6 Italian centres (Trieste, Milan, Parma, Rome, Naples, Bari) between 1973 and 1979, with the financial support of the Italian National Research Council. The results of the MPPI and other surveys are compared and the association between maternal smoking habits and sociodemographic background is investigated by multiple correspondence analysis. As to Italy, unlike UK and US, in the 1970s women of high social status show the highest prevalence of the smoking habit. Moreover, in pregnancy, the large majority gives up smoking, or at least reduces it, mainly in high socioeconomic levels, so that the proportion of pregnant women who keep on smoking over 10 cigaretts per day is very low (0.5-3.8%) and poorly related to sociodemographic factors. Therefore, it seems unlikely that these may exert serious confounding effects on the relationships between smoking in pregnancy and perinatal outcome.