Objectives-To determine the social costs of providing a rural population with radiology services under three different systems: the existing system (a small x ray unit at the remote site and all other examinations at the nearest radiology department (the host site)); a teleradiology system (most examinations at the remote site and more advanced examinations at the host site); and all examinations at the host site. Design-Cost minimisation study. Setting-Primary health care in a remote community in Norway. Subjects-A randomly selected sample (n=597) of all patients (n=1793) having radiological examinations in 1993. Main outcome measures-Annual direct medical costs, direct non-medical (travel) costs, and indirect costs (lost production) of the three options. Results-After exclusion of costs common to the three systems the direct medical, direct non-medical, and indirect costs of the three options were, respectively, pound 9000, pound 51 000, and pound 31 500 (total pound 91 500) for the existing system; pound 108 000, pound 2000, and pound 13 500 (total pound 123 500) for the teleradiology option; and pound 0, pound 75 000, and pound 42 000 (pound 117 000 in total) for the ''all at host'' option. Sensitivity analyses indicated that the existing system is the least costly option except when lost leisure is valued as highly as lost production. Conclusion-The teleradiology option did not seem to be cost saving in the study community. Such systems, however, may be justified on the grounds of equity of access and quality of care.