Background: Setting priorities is important in health research given the limited resources available for research. Various guidelines exist to assist in the priority setting process; however, priority setting still faces significant challenges such as the clear ranking of identified priorities. The World Health Organization (WHO) proposed a Disability Adjusted Life Year (DALY)-based model to rank priorities by research area (basic, health systems and biomedical) by dividing the DALYs into ‘unavertable with existing interventions’, ‘avertable with improved efficiency’ and ‘avertable with existing but non-cost-effective interventions’, respectively. However, the model has conceptual flaws and no clear methodology for its construction. Therefore, the aim of this paper was to amend the model to address these flaws, and develop a clear methodology by using tuberculosis in South Africa as a worked example. Methods: An amended model was constructed to represent total DALYs as the product of DALYs per person and absolute burden of disease. These figures were calculated for all countries from WHO datasets. The lowest figures achieved by any country were assumed to represent ‘unavertable with existing interventions’ if extrapolated to South Africa. The ratio of ‘cost per patient treated’ (adjusted for purchasing power and outcome weighted) between South Africa and the best country was used to calculate the ‘avertable with improved efficiency section’. Finally, ‘avertable with existing but non-cost-effective interventions’ was calculated using Disease Control Priorities Project efficacy data, and the ratio between the best intervention and South Africa’s current intervention, irrespective of cost. Results: The amended model shows that South Africa has a tuberculosis burden of 1,009,837.3 DALYs; 0.009% of DALYs are unavertable with existing interventions and 96.3% of DALYs could be averted with improvements in efficiency. Of the remaining DALYs, a further 56.9% could be averted with existing but non-cost-effective interventions. Conclusions: The amended model was successfully constructed using limited data sources. The generalizability of the data used is the main limitation of the model. More complex formulas are required to deal with such potential confounding variables; however, the results act as starting point for development of a more robust model.