Background Preterm birth-related complications are the leading cause of death in newborns and children under the age of 5 years. Maternal heat exposure has been associated with both sleep status during pregnancy and the increased risk of preterm birth. However, whether sleep status could mediate the association between heat exposure and preterm birth remains unclear. Objective To evaluate the association between maternal heat exposure in early pregnancy and preterm birth, and to further explore potential mediation effect of sleep status on the association between heat exposure and preterm birth. Methods A birth cohort was established in Guangzhou Panyu Maternal Child Health Hospital (Guangzhou Panyu District He Xian Memorial Hospital) from 2017 until now. Pregnant women (with gestational age between 8 and 13 weeks) were included in this study when they presented to the hospital for their first prenatal care visit and signed an informed consent. Then they were followed up until delivery. A total of 3 268 pregnant women were included for the final analysis. Questionnaires were distributed to collect the demographic characteristics, lifestyles, and sleep status of pregnant women. Daily meteorological data during the study period were collected from meteorological monitoring stations in Guangzhou and the average ambient mean temperature of four weeks before the survey was calculated and assigned for each pregnancy. The 75th, 80th, 85th, 90th, and 95th percentiles (P₇₅, P₈₀, P₈₅, P₉₀, and P₉₅) of the average ambient temperature of all pregnant women were used as the thresholds to define heat exposure. Logistic regression was used to evaluate the effects of heat exposure in different definitions on preterm birth and sleep status (sleep duration, night sleep timing, and wake up timing). The mediation effects of sleep status on the relationship between heat exposure and preterm birth were also analyzed. Results Among all the included participants, 165 newborns were preterm births with an incidence rate of 5.0%. Heat exposures with thresholds of P₉₀ and P₉₅ increased the risk of preterm birth, with ORs (95%CIs) of 1.66 (1.04-2.57) and 1.90 (1.03-3.33), respectively (P<0.05). Heat exposures with thresholds of P₇₅, P₈₀, P₈₅, P₉₀, and P₉₅ decreased the sleep duration (<9 h vs. ≥9 h, control group: ≥9 h), and the ORs (95%CIs) were 1.51 (1.25-1.83), 1.44 (1.17-1.77), 1.35 (1.08-1.70), 1.43 (1.09-1.87), and 1.45 (1.00-2.13), respectively. Heat exposures with P₇₅ and P₈₀ thresholds resulted in earlier wake up timing (<8: 00 vs. ≥8: 00, control group: <8: 00), with ORs (95%CIs) of 0.77 (0.63-0.93) and 0.76(0.61-0.93), respectively. No significant association was observed between heat exposure and night sleep timing. The mediation analyses showed that under heat exposure with P₉₀ threshold, a statistically significant mediation effect was observed for sleep duration, and the proportion mediated was 6.07% (95%CI: 0.17%-25.00%) (P<0.05). No significant mediation effect was observed for night sleep timing and wake up timing. Conclusion An elevated risk of preterm birth after heat exposure in early pregnancy may be partly mediated through reducing sleep duration.