The significance of the atmospheric stability approach in models for estimating evapotranspiration is studied by considering the errors encountered when simply adopting neutral instead of actual stability conditions. Maize, cotton and wheat crops grown in 10 hectare fields (central Greece) were used to estimate actual and maximum evapotranspiration for both stability considerations, employing a model based on the equations of Shuttleworth & Wallace (1985). The required meteorological data were recorded in each field continuously, whereas the leaf area index and the plant height were measured weekly. Results show that when estimation of cumulative evapotranspiration losses is attempted, the choice of the atmospheric stability approach is of no importance (error ≤ 2%), as over- and under-estimation of evapotranspiration are counterbalanced when stable and unstable situations are (under the climatic conditions of experimentation) successively interchanged during the cultivating period. When, however, daily estimations are required, the actual stability conditions have to be taken into account, as the daily evapotranspiration rate estimated by regarding the atmospheric stability as neutral may deviate from the corresponding ‘actual’ rate by as much as 50 – 80%.