Background: We aimed to evaluate if there is association between hours of sleep and the risk of obesity among children and whether this association differs by sex. Methods: A secondary data analysis, using information of the Young Lives study, was conducted. The outcome was obesity, based on the BMI for age z-score; the exposure was child's sleep duration (reported by parents) categorized using the National Sleep Foundation guidelines, and as a numerical variable. Baseline and three follow-ups information were used to evaluate association, reporting relative risks (RRs), and 95% confidence intervals (CIs), as well as coefficients and 95% CI. Results: Data from 1949 children, baseline mean age 4.3 (standard deviation: 0.3) and 962 (49.5%) females, were analyzed. Short sleep duration was present in 26.0% (95% CI: 24.0-28.0) at baseline. After 9.6 years of follow-up, the incidence of obesity was 0.83 (95% CI: 0.70-0.98) per 100 person-years at risk. In multivariable model (n = 1579), there was no association between short sleep duration and obesity in the whole sample (p = 0.13); but the risk of obesity was lower among girls (n = 816; RR = 0.45; 95% CI: 0.21-0.96; p = 0.03) compared with boys (n = 763; RR = 1.43; 95% CI: 0.95-2.14; p = 0.09). On the contrary, each additional hour of sleep was associated with an increase of boy's BMI mean (0.05; 95% CI: 0.02-0.08; p < 0.001), but not among girls (-0.02; 95% CI: -0.05 to 0.01; p = 0.11). Conclusions: Our results evidenced a lower risk of obesity due to short sleep duration in girls, but not in boys. Each additional hour of sleep was associated with an increase of BMI in boys, but not in girls. Strategies are needed to guarantee adequate sleep duration in Peruvian children.