The behaviour of a horseradish peroxidase (HRP) amperometric biosensor in different reversed micellar systems has been studied in order to extract overall conclusions about the possibility of using such media for developing enzyme amperometric biosensors. The enzymatic reaction employed involves the catalytic reduction of 2-butanone peroxide using ferrocene as mediator. The characteristics of the organic solvent used to form the reversed micelles, which affect the enzyme activity and stability, have been explored regarding the biosensor performance. The dependence of the electrode response on the amount of water in the micelle, the applied potential and the mediator concentration has been evaluated using n-hexane, chloroform and ethyl acetate as organic solvents. Ethyl acetate has been demonstrated to be the most suitable solvent for biosensor performance for analytical purposes. A systematic study of the different variables affecting biosensor behaviour when the nature of the surfactant used as the emulsifying agent is varied has also been carried out. Triton X-405, Tween 40, Hyamine 3500 and dioctyl sulphosuccinate (AOT) have been considered. The best reproducibility of the HRP electrode measurements was obtained for emulsions formed with AOT; these reversed micelles were the most appropriate for developing a method for the determination of 2-butanone peroxide. A detection limit of 0.4 muM was obtained.