Viruses pose a major challenge to sweetpotato production in Tanzania. Use of cleaned-up, virus-tested seed vines distributed through a formal seed system is among the proposed strategies to address this challenge. However, virus-tested seed vines can get infected once in the field and it is not known how they will perform following several seasons of on farm propagation. We assessed the performance of virus-tested seed vines and farmer-sourced seed vines of a susceptible variety, Ejumula, and a relatively tolerant variety, Kabode, over five seasons to understand the trend in root yields, vine yields and virus incidences. The experiments were done in high and low virus pressure areas. The most prevalent viruses were sweet potato chlorotic stunt virus (SPCSV) followed by sweet potato feathery mottle virus (SPFMV) and sweet potato leaf curl virus (SPLCV), respectively. Both farmer-sourced and cleaned-up, virus-tested seed of cv. Ejumula were rapidly infected with SPCSV. The incidence of this virus on Ejumula’s farmer-sourced material at the high-virus-pressure area reached 100% by the second season. The incidences for all three viruses remained stable for cv. Kabode across the five seasons. Plants generated from cleaned-up, virus-tested seed had lower incidences for all viruses compared to those from farmer-sourced planting material. Virus-tested seed produced significantly higher root yields for cv. Ejumula in the high-virus-pressure site, with a gradual drop across the seasons. The findings show that regular replenishment of clean, virus-tested seed is more economical in high-virus-pressure areas and for more susceptible varieties like cv. Ejumula. They also indicate that farmers may be reluctant to invest in cleaned-up, virus-tested seed in cases where they have virus-tolerant varieties such as cv. Kabode due to lack of obvious virus effect on yields.