Abstract: Based on the long data series of six weather stations in North China, three temperature-based ET₀ estimating methods, namely the Hargreaves, the McCloud and the Thornthwaite, were evaluated by comparing their results with that of the FAO56-PM. Three indicators, namely mean deviation, correlation coefficient and t-statistics, were used to compare yearly and monthly series estimated by the FAO56-PM and the temperature-based methods. Results show that the Hargreaves agrees best with FAO56-PM, followed by the McCloud, and then the Thornthwaite. In terms of yearly value, the temperature-based methods significantly underestimate the FAO56-PM in most stations of North China, with the Hargreaves by 53.2～200.2 mm/year or 4.6%～15.1%, the Thornthwaite by 269.7～468.1 mm/year or 24.8%～35.8%, and the McCloud by 90.5～435.7 mm/year or 8.2%～40.5%. The agreement between the temperature-based and the FAO56-PM method varies with months, overestimating the latter in summer, and underestimating it in other seasons, particularly in winter. The Hargreaves reaches peak value in June, which agrees with that of the FAO56-PM. The Thornthwaite and the McCloud reach peak value in July, the month with the highest temperature, and apparently lag behind that of the FAO56-PM. The Hargreaves can be chosen preferably in North China where only air temperature is available.