OBJECTIVES: Patients with non-ST elevation myocardial infarction are managed based visual assessment of coronary angiography, which can be inaccurate and subjective. A randomized pilot trial has investigated an alternative approach using physiology-guided management with fractional flow reserve that may optimize outcomes. Objective was to estimate the cost-effectiveness of FFR-guided management vs. coronary angiography. METHODS: A trial-based economic evaluation was conducted (n=350). We adopted a UK National Health Service perspective. Part I of the analysis used raw, unadjusted costs and QALYs assembled using individual resource use and EQ5D responses from the RCT. Part II used statistical modelling to model the effect pathway of FFR by conditioning total costs and QALYs on the treatment decision (coronary artery bypass graft (CABG), medical therapy and percutaneous coronary intervention (PCI)). Results were then applied to treatment decision distributions following FFR or standard care management. Uncertainty in GLM coefficients, unit cost parameters and sampling were incorporated using bootstrapping and Monte Carlo methods. RESULTS: FFR reduced revascularization by PCI or CABG (OR 0.52; 95%CI: 0.28 – 0.94; p=0.022). Part I: FFR led to a mean cost savings (£8,253 vs. £8,603; difference: –£305 [SD=£519]). Likely drivers of cost savings were length of stay (-£331 [£342]) and index year health events (-£217 [£223]). However, low information size contributed to their large imprecision. Incremental QALYs were comparable (0.811 vs. 0.799; diff: 0.013 [0.023]). Part II: FFR led again to a mean cost savings (£8,328 vs. £8,532; difference: –£204 [£305]) and comparable incremental QALYs (0.801 vs. 0.806; diff: –0.005 [0.004]). The probability of cost-effectiveness remained comparable over common willingness-to-pay (70-75%). CONCLUSIONS: Early evidence suggests FFR may be cost saving and comparable in QALYs. However, considerable uncertainty persists. We also prescribe specific design considerations for a future trial, including information size suggestions, follow-up frequency and outcomes to aid modelling/lifetime extrapolation.