Summary.To assess the role of non-compliance as a cause of instability in patients on oral anticoagulant therapy, a follow-up study of stably and instably anticoagulated patients and of patients beginning oral anticoagulant therapy was performed. Compliance was assessed by pill counting and with the use of pill bottles in the cap of which a microprocessor registered the exact date and time of opening of the bottle. (In)stability of oral anticoagulant therapy was expressed as the number of INRs and as the time spent within the target range and by squared sigma. Subsequently, as a pilot study, a randomized intervention study in instable, non-compliant patients was performed in which these patients were or were not informed about the real nature of the cap of the pill bottle.Nineteen stable and 19 unstable patients and 30 patients beginning therapy were followed for 3 months or less until therapy stopped. Compliance was better in the previously stably anticoagulated patients, although differences were small. Poor compliance was not a major cause of instability in patients starting therapy. Although the pilot intervention study was too small to assess the role of the special pill bottle, it was shown that compliance can be positively affected.