The first aim of the present study was to assess the validity of self-reported information concerning previous cancer. The second aim was to investigate determinants for under- and over-reporting of cancer. The Malmö Diet and Cancer Study is a prospective cohort study including 28,098 subjects. A questionnaire assessed lifestyle and medical conditions, e.g. previous cancer (yes/no). Cancer follow-up was made using the Swedish Cancer Registry (SCR). Questionnaire information concerning previous cancer was considered as test, and SCR information as golden standard (any tumour in the SCR, and specific sites were investigated). The validity analysis assessed agreement, sensitivity and specificity. Under- and over-reporting were studied in relation to lifestyle and socio-demographic factors. Sensitivity of the questionnaire in relation to malignant tumours was 0.82. It was 0.53 when cancer in situ of the breast and uterine cervix together with benign lesions of the urinary bladder and the nervous system were included in the cancer definition. Under- and over-reporting (in relation to malignant tumours) were more common among women and old people. Obesity, size of household, place of birth, smoking, social participation, educational level, type of employment, alcohol consumption and poor well-being was associated with under- or over-reporting in relation to at least one of the studied endpoints in either men or women. We conclude that there was a considerable disagreement between self-reported information and cancer registry data.