We demonstrate a transnationally situated dialogue as a method to bring ethnographic and historical research in Brazil, East Africa (Kenya, Tanzania and Uganda), India, Russia and Spain into conversation to show three cancer epistemics sites (research, detection, and care access) where the politics of cancer epistemics are at play. First, in the field of research, we show how certain ways of knowing, and certain questions about and interests in cancer, are privileged over others. Using examples from Spain and East Africa, we highlight how a shift towards microbiological and high-technology research has outpriced many more locally grounded research agendas, ignoring questions of industrial and capital accountability in cancer aetiology. Second, we look at ways of making cancer visible, how knowledge is mobilised in cancer detection and screening, where and for whom. We discuss the increased individualisation of risk which is reframing cancer surveillance and therapeutic agendas. Using examples from India, Spain and Brazil, we demonstrate how the epistemics of cancer detection generate discourses of blame and responsibility at the individual level and accentuate existing inequities whilst simultaneously absorbing patients and their families into complex networks of surveillance. Lastly, we examine how the epistemics of cancer implicate the very possibilities of accessing cancer care, shaping care pathways and possibilities for patients. With ethnographic examples from India, Russia and Brazil, we demonstrate how an orientation towards the individual shifts attention away from the commercialisation of healthcare and dominance of logics of profit in therapeutics. Throughout the paper, we point towards what is holding these cancer discourses together and grapple with how the politics of cancer epistemics are at play across the globe, even if they appear to be taking many different forms. Our approach highlights how practices are mirrored in the framing, implementation, detection and care of cancer with far-reaching effects.