AIMS: To evaluate the Limulus amoebocyte lysate (LAL) assay for differentiating Gram positive from Gram negative peritonitis in patients receiving continuous ambulatory peritoneal dialysis (CAPD). METHODS: One hundred and six patients with suspected peritonitis were studied. LAL assay was performed by adding 0.1 ml of CAPD fluid to 0.1 ml of LAL reagent and incubating in a heating block for 60 minutes at 37 degrees C. The sensitivity of the reaction was determined by: (i) diluting endotoxin in distilled water and used (filter sterilised) peritoneal dialysis fluid; and (ii) diluting a broth culture of E coli used in peritoneal dialysis fluid. A positive LAL assay was defined as the constant stability of the clot through an inversion of 180 degrees. RESULTS: Compared with bacterial culture, the LAL assay had a sensitivity of 65% and a specificity of 98%. The sensitivity of microscopy compared with culture of Gram negative organisms was 76%; overall sensitivity of microscopy in comparison was 80%. CONCLUSIONS: The Gram stain was more sensitive than has previously been reported; the LAL assay was specific but insensitive for the diagnosis of CAPD peritonitis. There was a correlation between reduced leucocyte count and culture; this was reduced in cases from which Gram negative organisms had been isolated. It is recommended that laboratories evaluate their Gram stain procedure to improve its sensitivity because the LAL assay is not a satisfactory substitute.