A case-control study among white women in Los Angeles County was conducted to investigate etiologic factors that might explain the high rates of invasive cervical cancer among Latinas. Two hundred patients with invasive squamous cell carcinoma of the uterine cervix and matched (age, sex, preferred language, and neighborhood) controls were interviewed, 98 pairs in English and 102 pairs in Spanish. Seven factors were found to contribute independently and significantly (P less than .01) to risk, each after adjustment for the other six: years since last Pap smear, years of education (protective), frequency-years douching, pack-years of smoking, years of barrier contraceptive use (protective), number of sexual partners before age 20, and recognized episodes of genital warts. An eighth variable, interval in years between menarche and first intercourse, was the second variable to enter the stepwise logistic regression analysis but lost its statistical significance when sexual partners before age 20 entered the model. Together, these eight variables accounted for almost 99% of the risk. There were no significant interactions between any of these variables and age, language of interview, or birth in a Latin country. There was no increased risk associated with use of oral contraceptives, either before or after adjustment for the other significant factors. Compared to English-speaking controls, Spanish-speaking controls smoked less, douched less, had fewer sexual partners before 20, and had essentially the same average interval between menarche and first intercourse and the same average number of episodes of genital warts; however, they had had a longer interval since their last Pap smear, fewer years of barrier contraceptive use, and fewer years of education. Education, presumably a correlate of an inadequately measured etiologic risk factor (possibly papillomavirus infection), was responsible for the greatest difference in risk between the Spanish- and English-speaking cases.