In this study the laboratory and clinical performance of a chromogenic endotoxin assay for equine plasma was evaluated. The assay was sensitive (detection limit 3 ng LPS/L plasma), reproducible (within and between-assay CV at 50 ng LPS/L E. coli O111:B4 LPS standard addition was 5% and 7.5%, respectively), and not substantially affected by enhancement or inhibition phenomena (recovery of an in vitro spike was 75-125% in 80% of the samples). LPS added to whole blood was rapidly inactivated upon incubation at 37 degrees C but not at 0 degrees C. A recently developed blood collection tube for LPS testing was found suitable, i.e. LPS-free and providing non-contaminated samples. In 48 horses suffering from acute abdominal diseases requiring surgical treatment, LPS levels were significantly higher in platelet-rich plasma (PRP) than in platelet-poor plasma (PPP), and the proportional difference was related to the PRP platelet count (r = 0.52, p < 0.001, mean difference 48%, range 8-77%). LPS levels were also significantly higher in horses that died or were euthanized than in surviving horses (mean 16.5 and 7.1 ng/L PRP, respectively, p < 0.05). We conclude that LPS can be measured in equine plasma with picogram sensitivity and recommend the use of PRP instead of PPP for clinical LPS testing. For clinical use a decision limit for endotoxaemia of 5 ng LPS/L PRP appeared to be inadequate. Analysis at a higher cut-off level for endotoxaemia and the evaluation of clinical, pathological, and laboratory parameters would be more meaningful.