This study examines long-term surface temperature variability and climate extremes in the Dieng Plateau, Central Java, from 1991 to 2022. Despite its tropical location, the region’s unique high-altitude microclimate, with frequent frost events, has raised concerns for local agriculture, particularly potato farming. However, limited observational data has constrained in-depth assessments. To address this, we used bias-corrected ERA5 reanalysis data, calibrated using hourly observations from an Automatic Weather Station (AWS) installed in 2021. The analysis focused on climatological trends and temperature-related extreme indices following the Expert Team on Climate Change Detection and Indices (ETCCDI) framework. Our findings indicate seasonal patterns in diurnal temperatures, with JJA (June–August) exhibiting the greatest variability and the lowest night time temperatures, conditions favorable to frost formation. Among the extreme indices, warmest night temperatures (TNx) increased significantly at a rate of 0.017°C/year (p < 0.01), while coldest night temperatures (TNn) showed a slight but significant decline. The frequency of warm nights (TN90p) rose by 0.242 days/month, while cold nights (TN10p) decreased by 0.161 days/month. Meanwhile, trends for warm days (TX90p), cold days (TX10p), and cold spell duration (CSDI) were statistically insignificant. These results highlight the plateau’s sensitivity to night time warming and the potential risk of climate-driven shifts in frost occurrence. The combination of high-resolution reanalysis data and extreme indices offers valuable insight into microclimate behavior in tropical highlands, with direct implications for frost risk management and climate adaptation strategies in vulnerable agricultural zones.