Drought is a slowly evolving, multi-scale hazard whose intensity in humid and sub-humid environments is increasingly modulated by rising temperatures rather than precipitation deficits alone. This study evaluated drought severity and frequency across the Jos Plateau of north-central Nigeria over a 31-year record (1995–2025) using the Standardised Precipitation Evapotranspiration Index (SPEI) computed at timescales of one, three, six, and twelve months. Potential evapotranspiration (PET) was estimated by the Hargreaves-Samani method from observed minimum and maximum temperature data. Mann-Kendall trend analysis and Sen slope estimation were applied to quantify directional changes in temperature, rainfall, and drought indices. Results reveal a statistically significant warming trend in maximum temperature (Sen slope = +0.0198°C year⁻¹; Z = 2.75; p = 0.006) and mean temperature (+0.0163°C year⁻¹; p = 0.014), superimposed on an essentially stable precipitation regime (p = 0.892). Annual PET (mean = 1,657 mm) consistently exceeded mean annual rainfall (1,245 mm) by approximately 413 mm, establishing a persistent climatological deficit. Three distinct drought clusters were identified: 1995–1996 (extreme; SPEI-12 = −3.09), 2015–2016 (extreme to severe), and 2018 (severe to extreme; four consecutive drought months). The middle decade (2005–2014) was characterised by an absence of moderate or severe drought conditions, representing a transient pluvial phase. Drought frequency in the most recent period (2015–2025) returned to early-record levels (9.1% of months in moderate-to-extreme drought), a pattern consistent with temperature-driven amplification of evaporative demand. The findings underscore the necessity of incorporating PET in drought assessments for semi-arid highland environments under observed warming.