Abstract. Near-surface air temperature (SAT) over Greenland has important effects on mass balance of the ice sheet, but it is unclear which SAT datasets are reliable in the region. Here extensive in situ SAT measurements ( ∼ 1400 station-years) are used to assess monthly mean SAT from seven global reanalysis datasets, five gridded SAT analyses, one satellite retrieval and three dynamically downscaled reanalyses. Strengths and weaknesses of these products are identified, and their biases are found to vary by season and glaciological regime. MERRA2 reanalysis overall performs best with mean absolute error less than 2 • C in all months. Ice sheet-average annual mean SAT from different datasets are highly correlated in recent decades, but their 1901-2000 trends differ even in sign. Compared with the MERRA2 climatology combined with gridded SAT analysis anomalies, thirty-one earth system model historical runs from the CMIP5 archive reach ∼ 5 • C for the 1901-2000 average bias and have opposite trends for a number of sub-periods.