Abstract Agricultural research has to tackle complex questions such as the design of sustainable cropping systems. System experiments are innovative approaches to address this challenge and a framework to iteratively design annual cropping systems has been proposed by Debaeke et al. (2009). However, specificities of some other cropping systems are not considered. Orchards are complex perennial agroecosystems formed of grass and tree layers aiming at the production of fresh fruit that require specific design and management over space and time. To identify orchard specificities and adapt the design framework to such perennial systems, we used two case studies of orchards aiming at decreasing pesticide use in temperate (apple, system experiment) and tropical (citrus, on-farm network) fruit productions. Specificities to take into account in the design framework are: (1) the spatial heterogeneity of the orchard with grass and tree layers, and tree rows and alleys; (2) the succession and interrelations among a young unproductive and then a productive stage; (3) the permanency of the fruit-tree crop constraining the management of soil fertility, (4) ground cover and (5) pest control, especially for pests that complete their lifecycle in the orchard and can build up important populations or inoculum across years. This is especially true in tropical areas where there is no dormant season. (6) Conversely, the permanency of the orchard habitats facilitates the sowing, planting or conservation of plant assemblages (e.g., ground covers, lining hedgerows) to enhance conservation biocontrol and/or compete weeds, provided non-disruptive practices are applied. Because of their longevity, orchards contribute to foster both plant-mediated (e.g., bottom-up) and natural enemy-mediated (e.g., top-down) processes in the foodweb to avoid direct measures against pests. Interactions among the orchard life stages, spatial and functional dimensions and practices need to be explicitly considered to optimize the efficiency of the system as a whole. Using the generic framework proposed by Debaeke et al. (2009) to design annual cropping systems, our framework includes adaptations to account for orchard specificities: (i) Agronomic objectives have to be fixed for each orchard stage; (ii) The cultivar choice and the composition and spatial arrangement of plants within the orchard are key elements to provide the expected services in the long term. This entails to include an additional step of perennial spatial design; (iii) Within-time and −space interactions have to be considered in the decisional system; (iv) Evaluation has to consider all the orchard stages in the global impact or performance, to account for carry-over effects and possible ‘paybacks’ of a given stage or period to the orchard whole lifetime. Last, to handle such complex interactions, design needs knowledge from many stakeholders in the food system (growers, advisors, scientists etc.) and requires more and renewed interactions among those stakeholders in a co-design process.