This paper deals with the application of a soft computing technique to a medical project. The application consists in connecting a patient to an ophthalmologist situated in a hospital or to a specialized web center through a telecommunication network. The patient may be at home or at any other place agreed upon for that teleconsultation. The aim of the study concerns the evaluation of a patient's visual acuity, using web-based medical advice. Indeed an ophthalmologist actually requires specific equipment, available only in medical centers and used by specialized staff. This situation leads to limit the number of patients being examined by the physician and so to extend the waiting time necessary before you can get an appointment (you have to wait between 8 and 10 months at the moment in France). So, public health services have an interest in making these tests available to the patients, in schools, firms, at home and so on, the tests being implemented remotely by competent personnel thanks to modern means of telecommunication. In order to achieve this, a certain number of tests are made, during which optotypes (or letters of different sizes) are presented to the patient. We propose an algorithmic treatment of the binary answers, based on the use of fuzzy sets, in order to eliminate the different mistakes (voluntary or not) that can be made. This algorithm also allows to assess the quality of the examination by measuring the inaccuracy of the visual acuity. This work must be considered as a pioneering telemedicine project, and particularly as a way of developing expert systems able to generate medical advice without the physician's actual presence.