<p>Potato is a key cash and food security crop in the tropical highlands of Ethiopia, a region experiencing increasing climate variability and change. Assessing the potential impacts of future climate scenarios on potato production is essential to inform effective adaptation strategies. This study calibrated, evaluated, and applied the SUBSTOR-Potato model to simulate climate change impacts on potato yields across three agroecosystems in the northwestern Ethiopian highlands. Calibration used field, survey, and secondary data, with cultivar-specific genetic coefficients estimated through a Generalised Likelihood Uncertainty Estimation (GLUE) approach. Model performance was evaluated using independent datasets. Calibration results showed strong agreement between simulated and observed values of phenology (<i>R</i>² = 0.93; <i>d</i> = 0.81), biomass (<i>d</i> = 0.97), and tuber yield (<i>d</i> = 0.98). Evaluation confirmed robust model performance, with high agreement in simulating phenology (<i>R</i>² = 0.90; <i>d</i> = 0.82), biomass (<i>d</i> = 0.98), and yield (<i>d</i> = 0.95). Thirty-year simulations were conducted for baseline (1981–2010), near-term (2011–2040), and mid-century (2041–2070) periods under two emission scenarios. Results showed substantial spatial yield variability, with the highest yield gaps and projected gains observed in high-elevation agroecosystems (AES5). Future climate scenarios project increases in both temperature and rainfall, potentially enhancing productivity if adaptive management practices are implemented. The calibrated SUBSTOR-Potato model provides a robust basis for site-specific impact assessments, supporting climate-resilient agricultural planning for smallholder potato systems in vulnerable highland regions.</p>