In recent years, the expansion of evidence-based interventions such as methadone maintenance therapy (MMT) has been at the forefront of global HIV prevention efforts for people who inject drugs. Despite these investments, there is an absence of social scientific research which treats 'evidence' and 'intervention' as objects of inquiry. Drawing on the case of MMT in Kenya, this critical sociological research study explores how globally endorsed health strategies are 'translated' into local health solutions according to the specific circumstances of their implementation. An analysis of longitudinal qualitative interviews, ethnographic observations, and primary source texts makes visible multiple 'versions' of MMT which are enacted through its policy and knowledge negotiations, and in relation to its local practices of delivery. The relative acceptability of MMT in Kenya is closely linked to its social capital as a medication of addiction recovery, even while its incorporation into policy is promoted by a globally supported HIV prevention evidence 'base'. At the same time, its medicalisation and strategic renaming as medically assisted therapy hold space for its inscriptions as both HIV prevention and addiction treatment. The multiplicity of MMT is not merely a discursive effect. Knowledge affected by bodily interactions, for example, enacts particular realities of MMT, including those not presumed to be 'in translation'. What's more, several strategies of bio/ disciplinary control-from the architectural features of the clinic enabling surveillance and segregation, to the rigidly enforced timetable for dispensing methadone-perform the intervention as a technology of governmentality. The friction and flux of MMT are rendered particularly stark by instances in which the effects of such practices trouble its representation in 'official' narratives. By illustrating how MMT, and the evidence produced in relation to it, are 'things in the making', this study stands in contrast with research positioned in global frameworks of evidence-based intervention.