<p>This study examined spatial and temporal variations and developed models to assess the density and severity of road traffic accidents (RTAs) in Washington, D.C., using accident data from January 2017 to April 2020. Descriptive statistics revealed that accident severity was highest during the morning peak (Severity Index = 0.77), while accident frequency peaked between 12 and 5 AM, particularly at 3 AM (4.62 accidents/hour). Weekly accidents rose from Monday to Friday, and seasonally, the summer months showed the highest crash counts. Spatial analysis using severity-weighted Kernel Density Estimation identified high-risk hotspots in central neighborhoods, disproportionately affecting pedestrians and cyclists. Regression analyses compared a Classical Regression Model (CRM) with two spatial econometric models: the Spatial Lag Model (SLM) and Spatial Error Model (SEM). For accident density, the SLM achieved the best performance (<i>R</i>² = 0.837, RMSE = 4.67, MAPE = 22.7%), outperforming both CRM and SEM, while confirming significant spatial autocorrelation (<i>ρ = </i>0.40). Similarly, for accident severity, the SLM provided the strongest explanatory power (<i>R</i>² = 0.542, RMSE = 0.106, MAPE = 16.1%), with a robust lag dependence (<i>ρ = </i>0.55). Across models, road density, commercial activity, and population density were positively associated with crash risk, whereas higher per capita income reduced severity outcomes. These findings demonstrate the superiority of spatial econometric approaches in capturing accident clustering and provide actionable insights for policymakers to prioritize interventions across high-risk locations and time periods.</p>