A few experimental studies suggest that atmospheric pollutants could affect the endocrine system, and in particular stress hormones and the hypothalamic-hypophyseal-ovarian axis, which could in turn influence menstrual cycle function. We aimed to study the possible short-term effects of atmospheric pollutants on the length of the follicular and luteal phases and on the duration of the menstrual cycle in humans. To do so, from a nation-wide study on couples' fecundity, we recruited 184 women not using contraception who collected urine samples at least every other day during one menstrual cycle, from which a progesterone metabolite was assayed, allowing estimation of the duration of the follicular and luteal phases of the cycle. Atmospheric pollution (nitrogen dioxide and particulate matter with an aerodynamical diameter below 10?m, PM10) levels were estimated from a dispersion model with a 1-km resolution combined with permanent monitoring stations measurements, allowing to estimate exposures in the 30-day, 1-10 and 11-30-day periods before the start of the menstrual cycle. Regression models allowed to quantify the change in cycle duration associated with atmospheric pollutants and adjusted for potential confounders. Follicular phase duration increased on average by 0.7 day (95% confidence interval, CI, 0.2; 1.3) for each increase by 10?g/m3 in NO2 concentration averaged over the 30 days before the cycle and by 1.6 day (95% CI, 0.3; 2.9) for each increase by 10?g/m3 in PM10. There was no strong evidence of associations of exposures in this time window with luteal phase or with total menstrual cycle durations (p>0.2). Exposures in the 1-10 day period before the cycle start were also associated with increased follicular phase duration. This study is one of the first prospective studies to suggest short-term alterations in follicular phase duration following atmospheric pollutants exposure.