OBJECTIVES A health technology assessment (HTA) informing the recent NICE guidance regarding the use of erythropoiesis-stimulating agents (ESA) in cancer-treatment induced anaemia (CIA) identified uncertainty around the overall survival hazard ratio (OSHR). We investigated how the understanding of OS in CIA patients treated with ESAs has shaped over time and the effects of accumulating OS evidence on cost-effectiveness. In addition, the effects of narrowing inclusion criteria, by comparing the HTA results to a recent Cochrane review, were investigated.  METHODS CMA was applied to both HTA review and Cochrane review OS data to identify patterns in results; study results were accumulated by the year of publication. Annual OSHR results from the CMA were applied to an economic model developed in the HTA to calculate the cost-effectiveness of ESAs. RESULTS Precision of the OSHR estimate appeared to improve with additional evidence, but the true location of the estimate remained uncertain and the best estimate varied over time. Using the HTA CMA, results from 2001 and 2002 suggested survival benefits to using ESAs (0.77, 95% CI 0.60–0.98 and 0.78, 95% CI 0.65–0.93 respectively), with ESAs being cost-effective at a willingness to pay threshold of £30,000 per QALY for all values of the OSHR 95% CI. HTA CMA for all other years and all Cochrane CMA results did not suggest any significant effects of ESAs on OS. Cost-effectiveness results were therefore uncertain.  CONCLUSIONS Current evidence suggests we cannot reject the possibility of no difference in OS between patients receiving or not receiving ESAs, regardless of study inclusion criteria. However, there is also insufficient evidence to support such conclusions, particularly as earlier results from narrower inclusion criteria suggested some survival benefits. This analysis highlights the additional uncertainty of the current evidence base on cost-effectiveness analyses, which cannot be captured in standard sensitivity analyses.