Background Birthweight is strongly associated with infant mortality and is a major determinant of infant survival. Several factors such as maternal, environmental, clinical, and social factors influence birthweight, and these vary geographically, including across low, middle, and economically advanced countries. The aim of the study was to investigate the geographical modification of the effect of oxides of nitrogen exposure on birthweight adjusted for clinical and socio-demographic factors. Methods Data for the study was obtained from the Mother and Child in the Environment birth cohort study in Durban, South Africa. Pregnant females were selected from public sector antenatal clinics in low socioeconomic neighborhoods. Land use regression models were used to determine household level antenatal exposure to oxides of nitrogen (NOx). Newborns were assessed at the local hospitals by doctors or qualified midwives. Six hundred and seventy seven births were analysed, using the geoadditive model with Gaussian distribution and identity link function. Continuous covariates were modelled by thin-plate regression splines and smoothing was estimated by the generalized cross validation. Findings A total of 677 births were analyzed. The newborns in the cohort had mean birthweight of 3106·5g (standard deviation (SD):538·2g and maternal mean age was 26 years (SD: 5·7). The mean prenatal exposure to NOx was found 31·5µg/m3 with standard deviation of 8·6 µg/m3. A spatially modified NOx exposure-related effect on birthweight was found across two geographic regions in Durban. Prenatal exposure to NOx was also found to have a non-linear effect on the birthweight of infants. The geoadditive model adjusted for maternal age, weight gain, and body mass index showed significant non-linear influences. Conclusion The study suggested that incorporating spatial variability is important to understand and design appropriate policies to reduce air pollution in order to prevent risks associated with birthweight and child development.