Water treatment yield, defined as the ratio of treated water output to raw water input, is a critical performance indicator for infrastructure in developing nations. In Rwanda, systematic analysis of long-term yield trends to inform operational and capital planning has been limited. This study aimed to develop and evaluate a robust time-series forecasting model to analyse historical yield performance and project future improvements for the country's water treatment systems, providing a methodological framework for evidence-based asset management. A seasonal autoregressive integrated moving average (SARIMA) model was applied to a national-level monthly yield dataset. The model structure was $\text{SARIMA}(p,d,q)(P,D,Q)_s$, with parameters optimised via the Akaike Information Criterion. Forecast uncertainty was quantified using 95% prediction intervals. The validated model projected a significant positive trend in national average yield, with a forecast increase of approximately 18 percentage points over the evaluation period. Model diagnostics indicated robust performance, with all residuals within control limits and a mean absolute percentage error of 2.7%. The SARIMA modelling approach provides a statistically sound methodology for tracking and forecasting water treatment efficiency, demonstrating its utility as a planning tool for infrastructure management. Water utilities should adopt this time-series methodology for routine performance monitoring and integrate the forecasts into medium-term investment plans to prioritise interventions at underperforming facilities. water treatment yield, time-series forecasting, SARIMA, infrastructure performance, asset management This paper presents a novel application of SARIMA modelling to forecast national water treatment yield, generating a validated predictive tool that fills a gap in quantitative performance analysis for the sector.