Exposure to ambient particulate matter during pregnancy has been suggested as a risk factor for preterm birth. However results from limited epidemiologic studies have been inconclusive. Very few studies have been conducted in areas with high air pollution levels. We investigated the hypothesis that high level exposure to particulate matter with aerodynamic diameter no larger than 10μm (PM10) during pregnancy increases the risk of preterm birth. A birth cohort study was carried out between 2010 and 2012 in Lanzhou, China, including 8969 singleton live births with available information on daily PM10 levels from four monitoring stations, individual exposures during pregnancy were calculated using inverse-distance weighting based on both home and work addresses. Unconditional logistic regression modeling was used to examine the associations between PM10 exposure and risk of preterm birth and its clinical subtypes. Increased risk of very preterm birth was associated with exposure to PM10 during the last two months of pregnancy (OR, 1.07; 95%CI, 1.02-1.13 per 10μg/m(3) increase for last four weeks before delivery; 1.09; 1.02-1.15 for last six weeks before delivery; 1.10; 1.03-1.17 for last eight weeks before delivery). Compared to the U.S. National Ambient Air Quality Standard (150μg/m(3)), higher exposure level (≥150μg/m(3)) of PM10 during entire pregnancy was associated with an increased risk of preterm birth (1.48; 1.22-1.81) and the association was higher for medically indicated preterm birth (1.80, 1.24-2.62) during entire pregnancy and for very preterm during last 6weeks before delivery (2.03, 1.11-3.72). Our study supports the hypothesis that exposure to high levels of ambient PM10 increases the risk of preterm birth. Our study also suggests that the risk may vary by clinical subtypes of preterm birth and exposure time windows. Our findings are relevant for health policy makers from China and other regions with high levels of air pollution to facilitate the efforts of reducing air pollution level in order to protect public health. Copyright © 2014 Elsevier Ltd. All rights reserved.