. This study tested the sensitivity of obesity diagnosis in electronic health records (EHRs) using body mass index (BMI) classification and identified variables associated with obesity diagnosis. . Eligible children aged 2 to 18 years had a calculable BMI in 2017 and had at least 1 visit in 2016 and 2017. Sensitivity of clinical obesity diagnosis compared with children’s BMI percentile was calculated. Logistic regression was performed to determine variables associated with obesity diagnosis. . Analyses included 31 059 children with BMI at or above 95th percentile. Sensitivity of clinical obesity diagnosis was 35.81%. Clinical obesity diagnosis was more likely if the child had a well visit, had Medicaid insurance, was female, Hispanic or Black, had a chronic disease diagnosis, and saw a provider in a practice in an urban area or with academic affiliation. . Sensitivity of clinical obesity diagnosis in EHR is low. Clinical obesity diagnosis is associated with nonmodifiable child-specific factors but also modifiable practice-specific factors.