A simple instrument for self-assessment of quality of life (QL) in patients with cancer was elaborated using a linear analogue scale (LAS). The instrument was based on five questions, exploring different functional areas; the same questions were also addressed in a parallel format, where problems were seen from an opposite point of view (positive/negative). The LAS was given to 222 patients, for a total of 372 tests collected. Internal consistency was satisfactory (Cronbach's alpha = 0.75); QL score was significantly correlated to parameters of disease. Concordance between scales, as judged by comparison of parallel formats, was statistically significant but poor. A questionnaire was then elaborated with similar items, based on a categorical scale. A direct comparison between LAS and our questionnaire was made on a group of 41 patients. Internal consistency was poor for the LAS (alpha = 0.58) and good for the questionnaire (alpha = 0.93); Spearman's rank correlation coefficients were disappointing for the LAS and good for the questionnaire; the questionnaire was judged reliable in 82.9% of cases, the LAS in 29.3% only; the questionnaire score, and not the LAS score, was significantly correlated with PS and disease status. In conclusion, many patients appeared unable to correctly interpret the visual-analogue scale; the categorical scale was more immediate and correctly understood by the large majority of patients; the correlation between score and important parameters of QL was maintained, and internal consistency was excellent, indicating a satisfactory reliability of this instrument.