OBJECTIVE To evaluate objectively the colposcopic skills of resident physicians in an obstetrics and gynecology training program. METHODS One hundred and ten colposcopic examinations were performed in a nonstructured fashion by 16 resident physicians supervised by the full-time gynecology faculty. Data were collected prospectively and were classified according to the resident's ability at colposcopic grading of cervical lesions (accuracy) and their ability to recognize the presence or absence of dysplasia (sensitivity and specificity, respectively) when compared to the corresponding histological interpretations. In addition, results between three groups were compared-junior residents (PGY1 and PGY2), senior residents (PGY3), and chief residents (PGY4)-to analyze learning progression. RESULTS The numbers of evaluations performed by junior, senior, and chief residents were 43, 33, and 34, respectively. The overall accuracy, sensitivity, and specificity of the residents' colposcopic assessments were 31.8%, 96.7%, and 22%, respectively. There was no difference in any of these measures between the groups analyzed. CONCLUSIONS Residents correctly identified cervical dysplasia in the majority of cases but had difficulty in distinguishing between low-grade and high-grade cervical dysplasia and in differentiating dysplasia from benign cellular changes, such as metaplasia and inflammation. Colposcopic accuracy, sensitivity, and specificity did not seem to improve with increasing experience. Resident colposcopic skills may improve if more structured training is implemented, particularly if emphasis is placed on teaching differences between benign cellular changes and dysplastic lesions.