We explored the effect of a diagnosis of cancer on employment according to cancer type, education, occupation, age, gender, mother tongue (Swedish or Finnish), calendar time and hospital district. All 12,542 new cancer cases diagnosed in 1987-1988 and 1992-1993, aged 15-60 years at the time of the diagnosis were identified from the Finnish Cancer Registry. The employment rate of the cancer survivors 2-3 years after the diagnosis was only 9% lower than their gender- and age-matched referents. However, we found that education and occupation modified the effect of cancer on the employment; the difference between cancer survivors and their referents in the probability of being employed was greater in the lower than in the higher educational groups. A modifying effect of education on the probability of employment was found among people with cancer of the lung, stomach, rectum and cervix uteri and those with cancers of the nervous system.