Transport maintenance depots are critical infrastructure for road safety and economic efficiency. In Kenya, these systems face persistent operational risks, yet quantitative methodologies for evaluating systemic risk reduction strategies are underdeveloped in the regional literature. This study presents a methodological evaluation of a multilevel regression model designed to quantify risk reduction within Kenyan transport maintenance depot systems. The objective is to assess the model's efficacy in isolating depot-level and regional-level variance in safety outcomes. A hierarchical dataset was constructed from maintenance records, audit reports, and incident logs across multiple depots. A two-level random intercepts model was fitted, specified as $y_{ij} = \beta_{0} + \beta_{1}X_{ij} + u_{j} + e_{ij}$, where $i$ denotes observations and $j$ denotes depot regions. Parameter estimation used restricted maximum likelihood with robust standard errors. The model accounted for a significant proportion of variance, with approximately 31% of the total variance in incident frequency attributable to regional-level random effects ($u_j$). A key concrete result is that a one-unit increase in standardised procedural compliance score at the depot level was associated with a 0.45 reduction in incident rate (95% CI: 0.38 to 0.52). The multilevel regression approach provides a statistically robust methodological framework for evaluating risk in interconnected depot systems, effectively partitioning variance across operational hierarchies. Infrastructure managers should adopt hierarchical modelling to inform targeted interventions. Future research should integrate temporal effects to model risk progression. multilevel modelling, infrastructure risk, maintenance depots, regression analysis, safety management This paper introduces a novel application of multilevel regression for partitioning systemic risk in transport infrastructure, providing a validated methodological tool for engineering decision-makers in similar contexts.