The extent to which a doctor or health professional can make a valid assessment of a patient's quality of life, anxiety and depression was investigated in a series of cancer patients. Doctors and patients filled out the same forms, viz. the Karnofsky, Spitzer, Linear Analogue Self Assessment Scales and a series of simple scales designed for this study, at the same time. Correlations between the two sets of scores were poor, suggesting that the doctors could not accurately determine what the patients felt. A further study examining the reproducibility of these scales demonstrated considerable variability in results between different doctors. It is concluded that if a reliable and consistent method of measuring quality of life in cancer patients is required, it must come from the patients themselves and not from their doctors and nurses.