PURPOSE: Foundation years or internships are an important period for junior doctors to apply their knowledge and gain clinical competency. Experiences gained during the foundation years or internships are likely to inform newly qualified doctors' opinions about how they want to continue their career. We aimed to understand how medical doctors' internship experiences influence their career intention/decision. METHODS: We conducted qualitative evidence synthesis using meta-ethnography. We searched six electronic bibliographic databases for papers published between 2000-2020 and included papers exploring how foundation years or internship experiences shape doctors' career intention/decisions, including in relation to migration, public/private/dual practice preference, rural/urban preference, and specialty choice. We used the GRADE-CERQual framework to rate confidence in review findings. RESULTS: We examined 23 papers out of 6085 citations screened. We abstracted three high-level inter-related themes across 14 conceptual categories: (1) Deciding the personal best fit both clinically and in general (which option is 'more me'?) through hands-on and real-life experiences (2) Exploring, experiencing and witnessing workplace norms; and (3) Worrying about the future in terms of job market policies, future training and professional development opportunities. Confidence in findings varied but was rated high in 8 conceptual categories. CONCLUSIONS: Our meta-ethnographic review revealed a range of ways in which internship experience shapes medical doctors' career intentions/decisions allowing us to produce a broad conceptual model of this phenomenon. The results highlight the importance of ensuring sufficient, positive and inspiring clinical exposure, improving workplace environment, relationship and culture, refraining from undermining specific specialities and communicating contractual and job market policies early on to young doctors, in order to attract doctors to less popular specialties or work locations where they are most needed. We propose our conceptual model should be further tested in new research across a range of contexts.