Total sleep duration has been decreasing among children in the last decades. Short sleep duration (SSD) has been associated with deleterious health consequences, such as excess weight/obesity. Risk factors for SSD have already been studied among school-aged children and adolescents, but inconsistent results have been reported regarding possible gender differences. Studies reporting such relationships are scarce in preschoolers, despite the importance of this period for adopting healthy behaviour. We aimed to investigate factors associated with SSD in 3-year-old boys (n = 546) and girls (n = 482) in a French Mother–Child Cohort (EDEN Study). Children were born between 2003 and 2006 in two French university hospitals. Clinical examinations and parent self-reported questionnaires allowed us to collect sociodemographic (e.g. income, education, family situation, child-minding system), maternal [e.g. body mass index (BMI), parity, depression, breastfeeding duration] and child's characteristics (e.g. gender, birth weight, term, physical activity and TV viewing duration, food consumption, usual sleep time). Sleep duration/24-h period was calculated and SSD was defined as <12 h. Analyses were performed using logistic regression. The mean sleep duration was 12 h 35 ± 56 min, with 91% of the children napping. Patterns of risk factors associated with SSD differed according to gender. In addition to parental presence when falling asleep, short sleep duration was associated strongly positively with high BMI Z-score and TV viewing duration among boys and with familial home child-minding and lower scores on the ‘fruits and vegetables’ dietary pattern among girls. These results suggest either a patterning of parental behaviours that differs according to gender, or a gender-specific sleep physiology, or both.