The aim of this paper is to compare the seroprevalence of cytomegalovirus (CMV) in women in the Autonomous Region of Madrid (ARM) obtained in two different years (1993 and 1999), according to age and seroprevalence of a virus transmitted by the faecal-oral route (hepatitis A virus, HAV) and another virus of respiratory transmission (varicella zoster virus, VZV). A total of 1813 serum samples were studied, taken from females aged 2-40 in two different serosurveys which were representative of the general population in the ARM. Of these, 933 were taken in 1993, and 880 in 1999. In each survey the samples were distributed over six groups, according to age group (2-5, 6-10, 11-15, 16-20, 21-30 and 31-40 years). CMV- and VZV-specific IgG was tested by indirect ELISA (Dade-Behring, Germany); and HAV-IgG by ImX (Abbott, USA) in the 1993 samples, or by Vidas (BioMérieux, France) in the case of those taken in 1999. A significant age-related rise in CMV seroprevalence was observed in both serosurveys. The seroprevalence obtained was lower in all age groups in 1999 than in 1993. The differences were statistically significant in two age groups: 6-10 years old (43.7 vs. 56.7%) and 31-40 years old (79.1 vs. 90.3%). In the younger age groups concurrent seroprevalence of CMV and VZV was significant lower in 1999. In older age groups a significant decrease in concurrent seroprevalence of both CMV and HAV was also seen. Agreement between serological results for CMV-HAV, CMV-VZV and HAV-VVZ during the two time periods and in every age group was poor or fair (kappa index < or = 0.2 or between 0.21 and 0.4) in all age groups. To conclude, a change in CMV epidemiology seems to be taking place in Madrid. The increase in the proportion of CMV seronegative women of childbearing age may have some impact on the incidence of congenital diseases related to vertical transmission of CMV. Apparently, such a change, among children, could be related to a lower close contact transmission rate (as in VZV), and among adults to improvements in standards of public health (as in HAV). However, due to the poor or fair agreement between serological results for CMV-HAV, CMV-VZV and HAV-VVZ, other independent factors may affect the fall in CMV seroprevalence.