Serum CA 125 regression after cytoreductive surgery and during the first three courses of chemotherapy was studied in 60 ovarian cancer patients and compared to known prognostic factors. Various methods reported in the literature to calculate a CA 125 half-live value were compared. Using two exponential regression models (Van der Burg et al., 1988; Buller et al., 1991), mean half-lives in stage I-II patients after complete cytoreductive surgery were respectively 10.7 days (range: 5-23) and 9.8 days (range: 7-15). Within stage III-IV patients, a significant positive correlation was seen between survival and (a) stage III (P = 0.002), (b) residual tumour < or = 1 cm (P = 0.02), (c) CA 125 normalisation after three courses (P = 0.003) and (d) CA 125 half-life < or = 20 days (P = 0.02-0.004, depending on the method used for half-life calculation). The median survival times of patients with and without a CA 125 normalisation after three courses were 27 and 14 months respectively (P = 0.003). When using the model of Buller et al. patients with a CA 125 half-life < or = 20 days had a median survival of 28 months compared to a median survival of 19 months for patients with CA 125 half-lives > 20 days (P = 0.004). Half-life calculations only showed a significant correlation with survival, if pre-surgery CA 125 levels were used as a baseline. In a survival analysis using the Cox proportional hazards model, stage of disease was the most predictive variable for survival (P = 0.006). The only additional independent prognostic factor for survival was the CA 125 half-life calculated according to Buller [derived from the formula: CA 125 = exp. [i-s x (days after surgery)], in which i is the y-axis intercept and s is the slope of the CA 125 regression curve]. A CA 125 half-life < or = 20 days vs > 20 days calculated using this formula, provides an independent prognostic factor for survival in stage III-IV patients early in the course of therapy (P = 0.04).