Patients with cervical carcinoma FIGO stage IB and IIA (n = 167) treated with brachytherapy, radical hysterectomy, and pelvic lymphadenectomy at the University Hospital of Uppsala were evaluated, and a multivariate analysis was performed to reveal clinical and histopathological variables of predictive value of recurrence. The 5-year survival rate was 90%. Nineteen patients developed recurrent disease (11%), 15 of whom died. Patient age, clinical stage, type, and histologic grade of tumor did not indicate an increased risk of recurrence. However, multiparity (3 children or more; relative risk, RR = 4.6), lymph node metastases (RR = 6.4), tumor size (RR = 5.1), and residual carcinoma in the hysterectomy specimen (RR = 3.4) were important predictive indicators of recurrence. The median interval from initial treatment to the diagnosis of recurrence was 15 months. The majority of recurrences occurred during the first 2 years after treatment (74%) and most of them had symptoms (84%). Only three patients with recurrence were diagnosed within our surveillance program. The data suggest that surveillance for recurrence can be made more cost efficient with a more individualized follow-up during the first 2 years after treatment, concentrating on the patients with high-risk factors such as large tumors, residual carcinoma after irradiation therapy, and/or lymph node metastases. (C) 1994 Academic Press, Inc.