The evaluation of telemedicine involves attempts to answer a wide range of questions involved in making decisions about safety, about practicality and about utility. Roughly speaking, if we wish to provide a telemedicine service we should first establish that it is safe, next that it is practical and finally that it is worthwhile. In establishing safety, most laboratory studies of telemedicine have a common structure, and consist of the following steps: (1) selection of cases; (2) interpretation; (3) comparison with a gold standard; (4) statistical analyses. Most of the studies to establish the practicality of telemedicine have been carried out as demonstrations, to show that a proposed application can be implemented in a chosen setting. In terms of utility, telemedicine has been used to improve the efficiency of an existing service or to make an existing service available to a new community. One of the difficulties is that the vendors of relatively expensive telemedicine systems and services disseminate much of the information on the topic. We have to focus not on the glamorous technology but on the underlying issue of how the participants in health care (patients, general practitioners, specialists) can communicate more effectively, using the range of technological options open to them. Ensuring that the most appropriate technology is used in the most effective way should be the primary aim of telemedicine research. There is now sufficient evidence for us to be confident that telemedicine is a safe alternative to conventional care in a variety of situations and for a number of clinical conditions. Reliable evidence that it is a practical and cost-effective alternative is, at the time of writing, harder to find.