Abstract Introduction Anticoagulation in cardioversion for atrial fibrillation is performed using unfractionated heparin and oral anticoagulants. TEE-guided cardioversion, after achievement of therapeutic anticoagulation (1–3 days), may be an alternative to the traditional procedure (3-week anticoagulation followed by cardioversion). The quality of anticoagulation in atrial fibrillation has not been investigated in a randomised trial with TEE-guided cardioversion. We analysed respective data from the ACE trial on the quality of conventional anticoagulation, where most participating centres chose the TEE-guided approach. Materials and methods In a randomised, prospective, multicentre trial, we analysed the efficacy of unfractionated heparin plus phenprocoumon in 248 patients on an intention-to-treat basis. There were 2373 evaluable anticoagulation measurements (out of 2925 measurements) and 4 categories of anticoagulation quality (under-, target, over- and severe over-anticoagulation). Of patients with evaluable measurements, 88% received short-term anticoagulation (4 weeks) in TEE-guided cardioversion. Results The median time to achieve therapeutic anticoagulation (aPTT ≥ 60 and < 80 s or INR ≥ 2 and < 3) was 3 days. Anticoagulation values were out of therapeutic range in 69.5% of measurements during 4- or 7-week follow-up, and never within therapeutic range in 10% of patients. Of the 15 primary endpoints observed (death, thromboembolism and major bleeding complications), only 3 were in patients with anticoagulation measurements within therapeutic range. Conclusions In this study setting, with predominance of 4 weeks anticoagulation in TEE-guided cardioversion for atrial fibrillation, therapeutic anticoagulation was reached within 3 days using conventional anticoagulation. Despite careful dose adjustments, anticoagulation was out of therapeutic range in almost 70% of total measurements and 80% of primary endpoints.