Since 2014, the Agricultural Operator Exposure Model (AOEM) has been the harmonised European model used for estimating non-dietary operator exposure to pesticide. It is based on studies conducted by the pesticide companies and it features 13 different crops including non-agricultural areas such as amenity grasslands. The objective of this study was to compare the dermal exposure measured during a field study conducted in a non-agricultural area with the corresponding values estimated by the model AOEM. The non-controlled field study was conducted in France in 2011 and included 24 private and public gardeners who apply glyphosate with knapsack sprayers. Dermal exposure was measured using the whole-body method and cotton gloves. Each measured value had an estimated value given by AOEM and we tested their correlation using linear regression. The model overestimated body exposure for all observations and there was no correlation between values. However, it underestimated hand exposure by 42 times and it systematically underestimated the exposure when the operators were wearing gloves, especially during the application. The model failed at being conservative regarding hand exposure and highly overestimated the protection afforded by the gloves. At a time of glyphosate renewed approval in Europe, non-controlled field studies conducted by academics are needed to improve AOEM model, especially in the non-agricultural sector. Indeed, among the 34 studies included in the model, none were conducted on a non-agricultural area and only four assessed the exposure when using a knapsack sprayer. Moreover, knapsack sprayers being the main equipment used worldwide in both agricultural and non-agricultural settings, it is also crucial to integrate new data specific to this equipment in the model. Operator exposure should be estimated with accuracy in the registration process of pesticides to ensure proper safety as well as in epidemiological studies to improve exposure assessment.< Réduire