A cross-sectional survey was conducted from August to October 1998 in CSWs attendingrnSTD clinics of two health centers in Addis Ababa, Ethiopia. The objective of the study was torndescribe socia-demographic characteristics, behaviors ofCSWs associated with HIV-Irninfection. In addition, a sub-group of CSW who remained uninfected despite long years of sexrnwork was identified and described in more detail. Purposive sampling was used for the survey.rnAll CSWs willing to participate were interviewed by female nurses. Blood samples were takenrnfrom each respondent for HIV and syphilis test. Characteristics of CSWs were comparedrnbetween groups using Chi-square and t test where appropriate. Identification of risk factors forrnHIV infection among CSW was performed using univariate and multivariate analysis (logisticrnregression model) A total of312 CSWs were enrolled in the study. Their mean age was 25rnyears. Most of the study participants had little education (786% were less than grade 6). Thernmean duration of sex work was 5.5 years. Their workplaces included individually ownedrnrooms (41.7%), followed by shared rooms (32,5%). Most of participants had already gotrninstructions about safer sex while working as CSWs, and 43 .0% ofCSWs reported using malerncondoms in at least 95% of sexual acts with clients. In multivariate analysis, variablesrnindependently associated with a protective effect against HIV infection were: higher educationrn(odds ofHIV infection associated with an increase of one category of education: 0.75, 95% CI,rn0.58-0.96), instruction by peers about sex work (OR 047,95% CI, 0.23- 0.96) and cleaningrnafter sex with soap (OR, 049, 95% cr, 0.25-0.94). Variables independently associated with anrnincrease in risk of HI V infection were: age 25-29 years (OR, reference category 15-19 years,rn3.24,95% CI, 106-9.95), sharing rooms with other sex workers (OR, 3.33, 95%CI, 130-8.5 1),rngenital bleeding after sex (OR, 2.34, 95% CI, 101-541), and history of genital ulcer in the pastrnyears (OR, 2.89, 95% CI, 1.39-5.99). There was no association between number of years ofrnsex work and HIV serostatus. Twenty five sex workers had more than five years of sex workrnand remained HIV negative. HIV prevalence was remarkably high (80.3%) among sex workersrnin Addis Ababa. Based on the results of this study, we recommend the strengthening of peerrneducation programs and of early diagnosis and treatment of STDs. The protective effect ofrncleansing using soap after sex should be explored further On the long run, creating alternativernemployment opportunities to sex workers should be strongly supported.