Accumulating evidence suggests that short- and long-term exposure to ambient fine particulate matter ≤ 2.5 μm (PM2.5) during pregnancy is associated with preterm births, yet the results are inconsistent, and the shape of the exposure-response curve is unclear, partially due to the limited studies conducted in areas with high air pollution. Our study evaluated the association between ambient PM2.5 concentration and preterm births in Beijing, China. Daily preterm birth data were collected from a hospital in Beijing during 2006 to 2013; a time-series of daily PM2.5 concentrations during the same period is assembled with measured data at three monitoring sites in Beijing. An extension of the Poisson regression and a time-series model were applied to simultaneously estimate the acute and chronic effects of exposure to PM2.5, with mutual adjustment for short- and long-term exposure as well as for confounders. During the study period, the PM2.5 concentration was 70.4 ± 60.6 μg/m3 and was found to be associated with an increased risk of preterm birth. In the study cohort, a 0.52% (95% confidence interval, CI: 0.09%, 0.96%) and 3.13% (95%CI: 1.92%, 4.35%) increase in preterm births was estimated for each 10-μg/m3 increase in short- and long-term exposure, respectively. This association was significantly modified by season (p < 0.05). With mutual adjustments for short- and long-term exposure, a more robust association (3.16%, 95% CI: 1.95%, 4.39%; per 10-μg/m3 increment in PM2.5) was observed for chronic effects. The exposure-response relationships for both short- and long-term exposure were linear, without a threshold, over the relatively low exposure range and flattened out at higher concentration levels. The maximum effect for long-term exposure to PM2.5 (33.6%) was much greater than that for short-term exposure (19.9%). These findings indicate that air quality improvements over a long period could yield significant health benefits.