We evaluated the feasibility and performance of visual inspection with acetic acid (VIA) and Lugol's iodine (VILI) for cervical cancer screening in a primary health-care setting in Kinshasa, Congo. Women (1,528) aged > or =30 years were screened independently by nurses and physicians by VIA and VILI and Pap cytology. Biopsy samples were obtained from women with abnormal colposcopies and from 290 randomly chosen women with normal colposcopy. Cytological and histological examinations were performed in Lyon and Montreal, respectively. The prevalence of cervical intraepithelial neoplasia (CIN) of grades 1, 2 and 3 was 4.5, 1.3 and 4%, respectively. Using biopsy as the reference, the sensitivity, specificity and negative predictive value (NPV) for > or =CIN 2 for VIA-nurse were 55.5% (95% CI: 34.7-76.2), 64.6% (95% CI: 62.0-67.1) and 96.8% (95% CI: 93.5-98.7), respectively. The corresponding values for VILI-nurse were 44.0% (95% CI: 24.2-63.8), 74.6% (95% CI: 72.3-76.9) and 96.7% (95% CI: 93.7-98.6). The equivalent parameters for physicians were 71.1% (95% CI: 46.7-95.5), 71.3% (95% CI: 68.9-73.6) and 98.6% (95% CI: 96.0-99.7) for VIA and 68.3% (95% CI: 42.5-94.0), 76.2% (95% CI: 74.0-78.4) and 97.2% (95% CI: 95.3-98.5) for VILI. The sensitivity of cytology ranged between 31 and 72%, depending on the abnormality threshold used to define positivity, with a corresponding specificity range of 94-99% and a NPV range of 97-99%. Our results show that VIA and VILI performed by nurses and physicians are slightly more sensitive but less specific than Pap cytology across multiple combinations of test and lesion thresholds. Given their lower cost and easy deployment, visual inspection methods merit further assessment as cervical cancer screening methods for low-resource countries.