The purpose of this study was to find out whether a decision-support system is able to assist a clinician in predicting patient outcome and in selecting optimal treatment in oncology. The domain of the evaluated decision-support prototype was primary therapeutic decision making in inoperable non-small cell lung cancer. The performance of the prototype was tested on retrospective material consisting of 112 patients treated by radiotherapy. Survival was the endpoint for examining whether the treatment decision proposed by the system was more accurate than the decision actually made by the clinician. Certain prognostic variables were used by the system to classify patients into two treatment groups, radical or palliative radiotherapy. The median survival times of these groups were 15 and 7 months, respectively, compared with 9 and 8 months in the corresponding groups classified by the clinician. Our results indicate that clinicians need support in treatment selection and that decision-support systems could be a potential answer.