School dropout remains one of the most persistent structural inequalities in Uganda's education system, and nowhere is this more acute than in the Karamoja sub-region. This study examined regional disparities in school dropout between Karamoja and national trends, drawing on a cross-sectional comparative design with a stratified random sample of 847 respondents, including students, teachers, parents, and local education officials, drawn from 32 schools across three Karamoja districts — Moroto, Kotido, and Napak — and a nationally representative comparison stratum of 18 districts across Uganda's four administrative regions. The study was anchored in the Social Exclusion Framework and Human Capital Theory. Descriptive analysis revealed that Karamoja's overall dropout rate stood at 57.6%, compared to 28.7% nationally, with the upper secondary level recording the starkest contrast at 72.4% versus 44.2%. Bivariate analysis using chi-square tests confirmed statistically significant associations between dropout and key determinants including poverty, gender, distance to school, cultural norms — particularly early marriage and cattle herding obligations — and teacher quality, all significant at p < 0.001. Structural Equation Modeling established that school absenteeism fully mediated the relationship between structural deprivation and dropout in Karamoja (β = 0.53, p < 0.001), with the Karamoja model explaining 67.4% of dropout variance compared to 53.1% nationally. Binary logistic regression further identified early marriage (OR = 6.83, p < 0.001), cattle herding obligations (OR = 4.57, p < 0.001), and poverty (OR = 3.84, p < 0.001) as the strongest predictors of dropout in Karamoja. The study concluded that Karamoja's dropout crisis is not merely a reflection of national trends but represents a structurally distinct and compounded educational emergency requiring targeted, context-sensitive policy intervention. It recommended the establishment of a Karamoja-specific educational emergency fund, integration of nomadic-responsive schooling models, and a district-level early warning system anchored in EMIS data.