AbstractNumerous epidemiological studies have shown that there is an association between smoking and cervical cancer. However, the essential evidence to show whether this relationship is casual or causal is lacking. The demonstration of DNA modification by tobacco components in the cervical epithelium would provide biochemical evidence to support a causal role. In this study, DNA from 39 cervical biopsies was analysed for the presence of DNA adducts using the 32P‐postlabelling technique. A questionnaire on smoking habit and a urinary cotinine assay were used to identify smokers and nonsmokers. DNA samples from smokers [identified from questionnaire] were found to have significantly higher adduct levels than nonsmokers (Mann‐Whitney one‐tailed U‐test, 95% CI > 0.339, P = 0.024). Exclusion of the women whose urinary cotinine levels did not confirm their self‐reported smoking status (smoker or nonsmoker) increased this significance (95% CI > 0.508, P = 0.01). Women who had abnormal cervical smears hadsignificantly higher DNA adduct levels than those with normal smears (95% CI > 0.439, P = 0.015). Monitoring of women with high DNA adduct levels may be a way of identifying women at risk of cervical cancer.These findings demonstrate that tobacco smoking by women leads to elevated levels of DNA adducts in cervical epithelium and provides the biochemical evidence to support the concept that smoking is a cause of cervical cancer. © 1994 Wiley‐Liss, Inc. Published 1994 Wiley‐Liss, Inc. © 1994 Wiley‐Liss, Inc.