This study aimed to identify the social, cultural and psychological characteristics influencing behaviour in a cancer mass screening campaign in a French population. The intention to take a screening test and actually doing it was studied, in particular. A self reported-questionnaire was mailed in December 1992 to a random sample of people living in Caen (western France), and aged from 45-74 years. The sample was formed by random selection from electoral registers. The study was population based, in the Caen area, department of Calvados, France. The questionnaire comprised 26 open and close questions. Starting in February 1993, the occult blood screening test (haemoccult IIR) for colorectal cancer was offered by general practitioners (GPs), occupational health doctors and pharmacists in the Caen area to all those aged 45-74 years. From 1 February 1993 to 30 June 1994, the data on the mass screening campaign were centralized and the study population divided into those who took the screening test and those who did not. Of the 1,129 persons contacted, 645 (57.1%) returned the questionnaire. After exclusions, 585 questionnaires were used for analysis. The results show that whether a person will actually take a screening test cannot be predicted from their intention to do so. The sociodemographic and cultural characteristics influencing the intention to take the test differ from those influencing execution of the test. Among the variables tested here, the following four sociodemographic and cultural characteristics were independently predictive of actually taking the screening test for colorectal cancer: compliance with the health insurer's advice; low or medium sociodemographic status, living with a partner (or widowhood); and not knowing someone with cancer. The analysis of the quantitative and qualitative differences between the intention to take the test and actually doing it could elucidate the reasons underlying refusal. Poor screening compliance has multiple causes. A record of intention to take a screening test alone is not appropriate. This type of study should no longer be carried out to determine the mechanisms underlying behaviour towards secondary prevention.