Non-invasive fibrosis tests (NITs) can be used to triage non-alcoholic fatty liver disease (NAFLD) patients at risk of advanced fibrosis (AF). We modelled and investigated the diagnostic accuracy and costs of a two-tier NIT approach in primary care (PC) to inform secondary care referrals (SCRs). A hypothetical cohort of 1,000 NAFLD patients with a 5% prevalence of AF was examined. Three referral strategies were modelled: refer all patients (Scenario 1), refer only patients with AF on NITs performed in PC (Scenario 2) and refer those with AF after biopsy (Scenario 3). Patients in Scenarios 1 and 2 would undergo sequential NITs if their initial NIT was indeterminate (FIB-4 followed by Fibroscan®, enhanced liver fibrosis (ELF)® or FibroTest®). The outcomes considered were true/false positives and true/false negatives with associated mortality, complications, treatment and follow-up depending on the care setting. Decision curve analysis was performed, which expressed the net benefit of different scenarios over a range of threshold probabilities (Pt). Sequential use of NITs provided lower SCR rates and greater cost savings compared to other scenarios over 5 years, with 90% of patients managed in PC and cost savings of over 40%. On decision curve analysis, FIB-4 plus ELF was marginally superior to FIB-4 plus Fibroscan at Pt ≥8% (1/12.5 referrals). Below this Pt, FIB-4 plus Fibroscan had greater net benefit. The net reduction in SCRs was similar for both sequential combinations. The sequential use of NITs in PC is an effective way to rationalize SCRs and is associated with significant cost savings. © 2019 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.