Two studies were sequentially conducted to validate a new questionnaire which takes into consideration the most important variables which could influence quality of life evaluation. Particular attention was given to the methodology employed to collect data and to the patients' characteristics. In the first study 80 consecutive cancer patients were randomised to twice fill in one of four different types of questionnaire, each one characterised by a different polarisation of semantic and syntactic extreme values of the visual linear analogue (for instance, "very much" always on the right, regardless of the semantic value of the answer; positive semantic value always on the right, regardless of whether it was "very much" or "not at all"; and so on). The second study, conducted on 60 lung cancer patients, consisted in testing the reliability (by measuring the reproducibility in different ways) and the validity (by performing a factor analysis) of the type of questionnaire indicated by the first study as the most reliable. The internal coherence was also evaluated by measuring the effects of physical and psychological conditions on the responses.