The value of serial CA-125 measurements for predicting progression of ovarian carcinoma during therapy was calculated in 71 patients. The optimal algorithm that defined disease progression by CA-125 levels was either two values above 100 U/ml which had decreased by less than 50% over a minimum of 56 days, or a rise of 25% between successive samples plus a confirmatory sample. Of 13 patients with progressive disease according to the CA-125 criteria, 12 developed clinical evidence of progression within 12 months; predictions were false positive in 1, true negative in 50 and false negative in 8. Retrospective analysis showed that therapy and investigations costing 7979 pounds could have been avoided, if CA-125 assays costing 5470 pounds had been acted upon. The efficacy of the CA-125 algorithm is being independently verified to confirm that monthly CA-125 measurements whilst on treatment combine cost-effectiveness with a decrease in unpleasant interventions.