Mental health services are increasingly encouraged to use co-design methodologies to engage individuals and families affected by mental health problems in service design and improvement. This scoping review aimed to identify research that used co-design methods with Culturally and Linguistically Diverse (CALD) communities in mental health services, and to identify methodological considerations for working with this population. In October 2019, we searched five electronic databases (CINAHL, PsycINFO, EMBASE, MEDLINE, Web of Science) to identify papers published in which people from CALD backgrounds were engaged in the co-design of a mental health service or program. Searches were limited to peer-reviewed articles published in English in the last 25 years (1993-2019). The search identified nine articles that matched the inclusion criteria. Using a scoping review methodology, the first author charted the data using extraction fields and then used qualitative synthesis methods to identify themes. Data were grouped into themes relevant to the research question. The two key themes relate first, to improving the experience for CALD communities when engaging in co-design research and second, to the development of co-design methods themselves. These findings support the need for further research into the transferability of co-design tools with CALD communities, particularly if co-design is to become a best practice method for service design and improvement. This scoping review identified methodological and practical consideration for researchers looking to use co-design with CALD communities for mental health service design, re-design or quality improvement initiatives. Further research is required to explore experiences of co-design methods, including documented protocols such as experience-based co-design, with CALD communities. This review indicates that explanatory models of mental health, community and co-design impact partnerships with CALD communities, and need to be understood to optimise the quality of these relationships when using co-design methods.