One hundred eighty patients with microinvasive and occult invasive squamous cell carcinoma of the cervix who were examined by colposcopy during a ten-year period are reviewed. Forty-two percent of patients with microinvasive carcinoma and 28% of patients with occult invasive carcinoma had an unsatisfactory colposcopic examination. Colposcopy led to the correct management in 90% of patients with occult invasive cancer and in 84% of patients with microinvasive carcinoma. Colposcopy appeared to be less sensitive in detecting microinvasive lesions than in detecting occult carcinoma. Two-thirds of the errors made were colposcopist-related, either failure of interpretation or protocol violation. A small but definite group of patients will have lesions whose colposcopic appearance is not sufficiently distinct or characteristic to permit a diagnosis of early invasion.