In the analysis of cost-effectiveness data from a randomized controlled trial, a confidence interval (CI) for the incremental cost-effectiveness ratio (ICER) is often presented to estimate the precision of the point estimate. Researchers have proposed a variety of approaches to construct such an interval. Fieller s interval is known to retain the coverage probability when the underlying distributions are normal. OBJECTIVE: To advocate the use of Fieller s interval and its extensions in estimating the CI for the ICER. METHODS: Use empirical adjustment to the modified signed log-likelihood ratio in situations where the underlying distributions are known and non-normal, and use the bootstrap-Fieller interval in situations where the underlying distributions are unknown. RESULTS: These methods were applied to data from the TRAndolapril Cardiac Evaluation (TRACE) study. CONCLUSION: These three likelihood-based methods are recommended to estimate the CI for the ICER to retain the coverage probability.