The prognostic value of immunosuppressive acidic protein (IAP), which is known to suppress various immune responses in cancer patients, was studied in a prospective randomized trial of advanced gastric cancer patients, designed to evaluate the effect of PSK, a kind of biological response modifier with protein-bound polysaccharides. Preoperative serum IAP levels were determined in 228 patients who received radical gastric resection and tests conducted in one laboratory by the single radial immunodiffusion (SRID) method. All patients were followed up for 24 months or more. There was an overall significant difference in disease-free survival time in favour of the PSK-treated group compared with the control group. Preoperative IAP values were strongly associated with disease-free survival time. The statistical analysis to define an appropriate cut-off level for IAP was performed using Cox's proportional hazards model. The most significant difference was observed at the threshold value of 580 micrograms/ml, the hazard ratio being 2.13 with a 95% confidence interval [1.17, 3.88] (P = 0.013). Patients in the PSK-treated group with a preoperative IAP of lower than 580 micrograms/ml showed improved disease-free survival (P = 0.029), however, no significant difference was seen between the two groups when the preoperative IAP exceeded the threshold level. From these results, 580 micrograms/ml is postulated to be the most appropriate threshold value for predicting the prognosis of advanced gastric cancer patients, and it is suggested that PSK would be most effective in patients whose preoperative IAP level is lower than the threshold level.