Although reanalysis products cover various variables on a global scale, their performance is different across regions. This study evaluates two reanalysis datasets, namely CAMS and ERA5, based on an in situ measured temperature over Tehran for three successive years of 2016, 2017, and 2018. The results show that both reanalysis datasets capture the general pattern of daily and monthly variability. Both datasets contain a cold bias. Despite the enhanced spatial resolution of ERA5, it turns out that ERA5 is more biased than CAMS. Furthermore, a detailed assessment of model performance indicates that more than 80% of the error is associated with bias.