We evaluated associations of community types and features with new-onset internalizing disorders among Pennsylvania adolescents to identify the location and scale of risk. Using a nested case-control study, we drew subjects from electronic health records 2008–2016, requiring cases (n = 7974) to have two medication orders or diagnoses indicating an internalizing disorder; controls (n = 31,895) were frequency-matched. Subjects were assigned to three community classifications: townships, boroughs, city census tracts; urbanized areas, urban clusters, rural areas; and a combination. Using logistic regression with generalized estimating equations, we found that compared to rural-townships, the highest odds were in urban cluster-city census tracts (odds ratio, 95% confidence interval: 1.78, 1.41–2.26); lowest in urbanized area-city census tracts (0.85, 0.74–0.97). Higher community socioeconomic deprivation was associated with increased odds in urban clusters (1.21, 1.00–1.48) and higher greenness with decreased odds in urban clusters (0.73, 0.62–0.86).