Epidemiological studies which examine an association between low serum cholesterol level and cancer risk of all sites were reviewed and a summary of major findings are as follows: (1) In most reports, low serum cholesterol level, which associated with cancer of all sites, cancer deaths/incidence occurred within a few years after cholesterol measurement were performed. This short-term association is believed to be a preclinical cancer effect, i.e., preclinical cancer itself reduces serum cholesterol level. (2) A long-term inverse association between serum cholesterol and cancer risk of all sites was observed in some studies. This association cannot be explained by either a preclinical cancer effect or by chance. The epidemiological implications of this association was discussed in depth. (3) A long-term inverse association between serum cholesterol level and cancer mortality/incidence was observed mainly in men, but not in women. In a few studies which reported an inverse association in women, follow-up periods were rather short. (4) A few epidemiological studies conducted in Japan also detected a long-term inverse relationship between serum cholesterol level and cancer risk of all sites in men, but not in women. Since low serum cholesterol level may possibly be a risk factor for cancer of all sites, further investigation involving experimental studies on animals and longitudinal population-based study appear to be necessary for confirmation of this factor.