Objective: To compare all-cause healthcare costs among non-valvular atrial fibrillation (NVAF) patients who switched from warfarin to novel oral anticoagulants (NOAC). Methods: Adult NVAF patients who switched from warfarin to dabigatran, rivaroxaban or apixaban were identified in MarketScan claims databases between 10/01/2010 and 12/31/2015. Patients were continuously enrolled for 12 months before the first NOAC claim and followed for 12 months or until medication switch, discontinuation, inpatient death or 12/31/2016. Dabigatran patients were matched 1 : 1 separately to rivaroxaban and apixaban. All-cause costs were reported as per-patient-per-month (PPPM) in 2017 US dollars. Key findings: A total of 8679 and 5761 dabigatran switchers were matched to rivaroxaban and apixaban switchers respectively (mean age 73–74 years; mean CCI 1.8–2.0). Compared with rivaroxaban, dabigatran switchers had significantly lower PPPM mean outpatient (OP) ($1265 versus $1587, P < 0.001), emergency department (ED, $67 versus $95, P < 0.001), OP office ($114 versus $119, P = 0.003), other OP services ($1085 versus $1373, P < 0.001) and OP pharmacy costs ($624 versus $660, P < 0.001). Compared with apixaban, dabigatran switchers had significantly lower mean PPPM ED ($67 versus $123, P < 0.001), OP office ($116 versus $121, P = 0.032), other OP services ($1062 versus $1434, P < 0.001), OP pharmacy ($633 versus $706, P < 0.001) and total healthcare costs ($3254 versus $3805, P = 0.016). Conclusions: Outpatient costs were considerably lower among dabigatran switchers compared with rivaroxaban. Total and OP healthcare costs were significantly lower for patients switching from warfarin to dabigatran versus apixaban. Use of dabigatran following warfarin discontinuation may enable healthcare cost savings among NVAF patients, as compared with rivaroxaban or apixaban.