{ "background": "Municipal infrastructure asset systems in Rwanda face challenges from rapid urbanisation and constrained maintenance resources. Current reliability assessments often fail to account for the hierarchical nature of asset data, where individual components are nested within larger systems and geographical units.", "purpose and objectives": "This study aimed to develop and apply a multilevel regression modelling framework to assess the reliability of municipal infrastructure systems, explicitly accounting for data hierarchy to provide more accurate and actionable insights for asset management.", "methodology": "A hierarchical dataset on water supply, roads, and public buildings was analysed. A two-level random intercept logistic model was specified: $\\logit(p{ij}) = \\beta{0} + \\beta X{ij} + u{j}$, where $p{ij}$ is the probability of asset $i$ in district $j$ being fully functional, $X{ij}$ are asset-level covariates, and $u_{j}$ are district-level random effects. Model parameters were estimated using restricted maximum likelihood.", "findings": "District-level random effects accounted for 31% of the variance in asset functionality, indicating significant geographical clustering. A one-unit increase in a standardised maintenance frequency index increased the odds of full functionality by 2.4 times (95% CI: 2.1 to 2.8).", "conclusion": "The multilevel model successfully captured substantial geographical heterogeneity in asset reliability, which conventional single-level models would overlook. This confirms the necessity of hierarchical modelling for infrastructure systems management.", "recommendations": "Municipal asset management plans should adopt hierarchical data collection and analysis frameworks. Resource allocation should be informed by both asset-specific conditions and the identified district-level effects to improve system-wide reliability.", "key words": "asset management, hierarchical modelling, infrastructure reliability, multilevel regression, urban services", "contribution statement": "This paper provides a novel methodological application of multilevel regression for infrastructure reliability assessment in a Rwandan context, demonstrating how accounting for data hierarchy yields superior insights for priorit