Managing pests while reducing pesticide use is a major challenge to improve cropping system sustainability. Crop diversification appears as a promising tool to promote ecosystem services such as pest control (Tamburini et al., 2020; Beillouin et al., 2021). Lechenet et al. (2016) identified crop sequence and crop diversity, among other management practices, as important factors to reduce pesticide use in most French production situations. Crop diversification could allow the reduction of pesticide use at the cropping system level through: i) the introduction of crops with low intrinsic pesticide use (i.e. grassland, sunflower,…) and ii) the disruption of biological cycles of pests and diseases in time and space. Nevertheless, the reduction of pesticide use allowed by crop diversification has never been quantified nor the stability of its effect highlighted across diverse production situations. We addressed this issue by using the Dephy network dataset, a national network of 3000 farms initiated as part of the French national plan Ecophyto, which aims to reduce pesticide use by half across the French territory. We considered 1614 non-organic cropping systems described annually over a period of one to thirteen years. Cropping systems displayed a wide diversity of management strategies (i.e. in the crop rotation and in the specific management of each crop) within and across various pedoclimatic conditions. For each cropping system and year, the number of crops and the return delay in the rotation of nine main crops (i.e. wheat, barley, oilseed rape, maize, potatoes, sugar beet, sunflower, pea and soybean) were computed. Pesticide use at the crop level was quantified with the Treatment Frequency Index (TFI). At the cropping system level, TFI was expressed as the weighted average of each crop’s TFI according to its proportion in the sequence. The relationship between the number of crops in the cropping system and TFI at the cropping system level was modelled as a second order polynomial, using a linear quantile mixed model. The effect of return delay of nine crops on pesticide use was quantified at the crop and cropping system level and modelled in interaction with climatic region, using generalized linear mixed models. Increasing the number of crops from one to five tended to increase cropping system TFI, because crops with low pesticide use (i.e. maize and meadow) tended to be replaced by crops with higher pesticide use. However, when the number of crops in the rotation was higher than five, cropping system TFI decreased, especially for the 25% cropping systems with the highest TFI. For five crops out of nine (i.e. wheat, oilseed rape, maize, sugar beet, and barley) increasing the return delay from one to five years reduced crop TFI from 0.3 (i.e. barley) to 1.2 points (i.e. wheat). In most cases, increasing the return delay of crops also scaled up to a reduction of TFI at the cropping system level, which could either reflect either a selection effect (substitution of crops strongly relying on pesticides (i.e potatoes) with crops with low intrinsic pesticide use) or a regulating effect of diversified crop rotations on pest pressure in a given crop. Conversely, for maize in the oceanic degraded climatic region and for sunflower in all climatic regions (i.e. two crops with low intrinsic pesticide use) increasing the return delay from one to five years induced a small increase of cropping system TFI (i.e. +0.6 and +0.2 point, respectively). Hence, increasing their return delay may lead to the introduction of crops with higher pesticide use. To go further, a more detailed characterization of crop sequences associated to low pesticide use is needed. Tamburini G., et al. (2020). Science advances, 6(45), eaba1715