There is a growing concern about the possible adverse effects of exposure to air pollution on health during pregnancy. Therefore, a priority of the INMA (environment and childhood) study was to estimate personal exposure to traffic-related air pollution. In the cohort from Valencia (n=855), ambient levels of NO(2) were measured at 93 sampling sites spread over the study area during four different sampling periods of 7 days each. Multiple regression models were used to map ambient NO(2) over the area. Geographical data and predictions from kriging obtained by the "let one out" procedure were used as predictors. Individual exposure was assigned as 1) the estimated ambient NO(2) level at the home address and 2) the average of estimated ambient NO(2) levels at home and work addresses, weighted by the time spent in each environment. Estimations were temporally customized using the NO(2) levels registered daily by the regional Air Pollution Monitoring Network. The entire pregnancy and each trimester were taken as exposure windows. The model for the mean levels of NO(2) during the sampling periods explained 81% of the variation in NO(2) levels. Relative percent differences between the two models of personal exposure assignment were less than 9% for more than 90% of the participants; however the rest of them showed marked differences. Personal exposure estimates were slightly higher in the second model. In both cases, exposure during the whole pregnancy was strongly correlated with exposure in the second trimester. Considering periods shorter than the entire pregnancy will provide us the opportunity to identify specific windows of susceptibility.