Abstract In Lake Zone, Tanzania, the Sweetpotato Action for Security and Health in Africa Marando Bora project trained farmers to become specialised multipliers to produce and distribute sweetpotato seed. The objectives of the case study were to analyse (1) how farmers and trained multipliers characterised and managed quality in sweetpotato planting material; (2) changes in skills and practices as multiplication became a specialised task; and (3) what other transactions were happening as planting material was exchanged. The study used mixed methods to investigate the sweetpotato seed system as a social and technical configuration and the processes and interactions within it. A formal survey of 88 trained decentralised vine multipliers (DVMs) was conducted in March 2013. Detailed observations and semi-structured interviews were also carried out with 21 farmers and a sub-group of the DVMs. The survey results showed that 70% of DVMs were multiplying vines 9 months after the project finished; 34% were using close spacing in beds; 61% were using conventional plant spacing on ridges for roots and vines; and 5% were using both multiplication beds and ridges. Further, 97% of DVMs irrigated from wells, springs, lake, river, or dam water; 34% used pumps. During the life of the project, 89% of DVMs used fertilizer, and 26% continued to use it after the project finished. The in-depth interviews showed that farmers selected planting material not only on the basis of plant health, but also used cues to plant vigour. The use of own-saved seed involved a complex performance of managing seed and root production across different agro-ecologies according to inter- and intra-season characteristics. As the vine multiplication cycle became a specialised activity, vines were treated differently, depending on whether they were to be used for high root production or high vine production. The resilience and viability of the existing seed systems depended not only on individual knowledge and skills, but also management practices across agro-ecologies and negotiating social settlements at community level. As specialisation of seed production developed, the multiplication and root production cycles were separated and tasks became segregated. New knowledge and skills were formed in different ways; the trained multipliers adapted the new techniques into the existing system of shifting vine conservation and root production between different ecologies. Interventions aimed at building the capacity of specialised vine multipliers and scaling-up seed interventions should consider the implications of skills-building and segregation of tasks in the broader context of society choice of technologies and agrarian change. Margaret A. McEwan