Food self-sufficiency (FSS) and healthy diets are high on policy agendas to ensure food security under increasing global pressures. A global shift towards self-sufficient production of healthy diets would represent a radical departure from today's globalised food system. Representing such scenarios in a biophysically consistent way requires accounting for multiple resource constraints and feedback loops—including feed, fertiliser, and trade flows—while allowing flexible reallocation of crop areas, livestock numbers, and biomass streams. We use the global biophysical optimisation model CiFoS (Circular Food Systems) to evaluate the potential for self-sufficient production of multiple food groups and nutrients in 70 regions by 2050 under a business-as-usual diet (BAU-MinTrade) and a Planetary Health Diet (PHD-MinTrade). FSS is assessed by minimising biomass and nutrient trade while fulfilling dietary requirements, with trade only balancing shortages. Results show that total trade could fall by 62% to 618 million tonnes in BAU-MinTrade and by 79% to 343 million tonnes in PHD-MinTrade. Many regions—including Europe, the Americas, Oceania, and China—could be almost self-sufficient under both scenarios. Several African regions, India, and parts of Asia would still rely on imports, especially under BAU-MinTrade. Most food groups and nutrients show potential for increased FSS, though trade in some animal-source products and nutrients may rise. Self-sufficient systems can keep land use and GHG emissions within planetary boundaries, but nitrogen and phosphorus inputs remain high. PHD self-sufficiency is consistently more sustainable than BAU. Aligning production with dietary shifts towards a PHD supports self-sufficiency while reducing environmental trade-offs.