The aim of the study was to assess the accuracy of fibronectin, a glycoprotein, for the diagnosis of malignant ascites and to compare it with conventional parameters. Ascitic fluid samples from 50 patients, 25 with intra-abdominal malignancy and 25 without it were analysed for total protein concentration, fluid/serum protein ratio, glucose concentration, leucocyte count, pH, fibronectin concentration (by ELISA) and for malignant cell cytology. Twenty-two of the 25 patients with ascites and intra-abdominal malignancy had documented peritoneal metastases in group A. The 25 patients with non-malignant ascites constituted group B. Mean values of ascitic fluid fibronectin, for groups A and B were 538 +/- 46 micrograms/mL and 60 +/- 4.92 micrograms/mL, respectively (P less than 0.001). Within the group with malignant ascites, patients who had positive malignant cytology (n = 12) exhibited a significantly higher ascitic fluid fibronectin concentration than patients with negative cytology (P less than 0.05). While mean ascitic fluid protein concentration showed a significant difference (P less than 0.01) between the two groups, there was no difference in respect to ascitic fluid pH, glucose concentration and leucocyte count. Malignant cell cytology was positive in 54.5% of group A patients with no false positive report in group B. The diagnostic accuracy for differentiating malignant from non-malignant ascites was 100% for a fibronectin value of greater than or equal to 110 micrograms/mL as compared with 78.7% for ascitic fluid protein concentration greater than or equal to 0.5 g/dL, 57.4% for leucocyte count greater than or equal to 1000/mm3, 59.6% for pH less than 7.45 and 78.7% for malignant cell cytology.