Preterm birth is the leading cause of neonatal and infant mortality in developed countries. The risk of preterm birth has been reported to vary by season in several countries, including Japan and the USA. Discovering the reasons for such variations may contribute toward our understanding of preterm birth aetiology and enable the development of more effective interventions. Using time-series data from the St Mary’s Maternity Information System (SMMIS) from 1988 to 2000, I have been able to demonstrate that the probability of preterm birth varies by season in northwest London, Hertfordshire and Bedfordshire. The incidence of preterm birth was consistently higher during winters than during summers, with a 10% increase in risk of being born preterm in winter when compared with summer (RR 1.10, 95% Cl 1.07 to 1.14). The seasonality appeared to be largely due to births that occurred later during the preterm period (i.e., 32 to less than 37 weeks of gestation) rather than earlier preterm births (i.e., 24 to less than 32 weeks of gestation). My primary hypothesis was that this seasonal variation might be explained by climatic effects, including air pollution. I also investigated the possibility that infections, known to be a cause of some cases of preterm labour, might vary in a way that was associated with the risk of preterm birth. Regression techniques were used to investigate whether the seasonal variation in preterm births was explained by any short-term associations with various meteorological, air pollution or infection factors. Other aspects of preterm birth, such as the type of clinical presentation and births necessitated by maternal pre-eclampsia were also investigated as potential mediators. Possible associations on the day of birth, cumulative effects from up to six weeks before birth and effects from exposure around the time of conception were investigated. To check for possible effect modification, each model was also stratified by maternal age, maternal ethnicity, sex of the fetus, gestational age of preterm birth, and parity. The findings from the analysis of short-term associations confirmed the complexity of the pattern of preterm birth. While associations with some meteorological factors, influenza and PM10 were found, these did not appear to explain the seasonal pattern of preterm birth proportions that was observed. Nor was the seasonal pattern of pre-eclampsia or pregnancy-induced hypertension strong enough to explain more than a small proportion of the seasonal pattern of medically indicated preterm births. My findings did support, however, the theory that there are different causative mechanisms between early and late preterm birth. I also found that the seasonal patterns and associations varied between different ethnic groups. My study has for the first time established a seasonality of preterm births in a British cohort and used a novel method for investigating and dissecting the multiple intersecting pathways that lead to preterm birth. In the current era, when climate change is widely predicted, it is important to study the impact this may have on our health. There is a need to develop methods to assess the potential impact of environmental factors on our well-being, and in particular, reproductive health. Further research related to understanding the mechanisms driving the seasonal pattern of preterm birth is warranted and could prove important in future efforts to prevent or reduce preterm birth and its related consequences.