At the global scale, urban agriculture is increasingly developing in cities due to demographic growth and sustainable food concerns. But, urban soils are frequently polluted with metals. In urban gardens, organic matter is also commonly added both to valorize organic household waste and to promote biophysicochemical fertility. As earthworms promote the decomposition and the recycling of soil organic matter, they can also influence the biogeochemical cycle of metals in urban polluted soils. In order to produce safe vegetables in urban areas, it is crucial to highlight the mechanisms involved in complex soil-earthworm-plant ecosystems. An experiment was set up to examine these relationships using lettuce cultivated in controlled conditions with RHIZOtest® devices. Thanks to the RHIZOtest® devices, metal transfer and bioaccessibility were for the first time compared for urban polluted soil without (1-urban soil polluted with Pb, Cd, Cu, and Zn: essential or toxic metals currently found in environment, SNB) and with bioturbation (2-this metal-polluted soil subjected to earthworm bioturbation, SB) and earthworm casts (3-earthworm casts produced in this polluted soil and naturally enriched in organic matter and microorganisms, T). Metal concentration, phytoavailability, and human gastric bioaccessibility were determined in the different samples. Results showed that earthworm bioturbation increased the phytoavailability of all the metals. For the experimental condition SB, the phytoavailability of metals was increased up to 75% compared to SNB. In addition, surprisingly, metal phytoavailability was always superior in SB compared to earthworm casts (T). Moreover, earthworms led to an increase in Zn gastric bioaccessibility up to 10% in the soils in the same way as for phytoavailability, meaning Zn bioaccessibility in SB > T > SNB, whereas it remained unchanged in the lettuces. These data are important to promote sustainable agriculture activities in urban areas; actually, databases concerning different experimental conditions are needed to develop decision support tools.