Serum amyloid A (SAA) in serum was the biomarker which best improved the clinical assessment in differentiating infectious colic from surgical colic. Introduction:The objective of this study was to investigate the ability of the acute phase proteins (APPs); Serum amyloid A, haptoglobin and fibrinogen to differentiate between horses with infectious non-surgical colic and surgical colic. Materials & Methods:The performance of the APPs was evaluated individually and in combination with clinical examination, as wells as traditional biomarkers in blood (WCC, PCV, TPP, lactate) and peritoneal fluid (PF) (haemolysis, WCC, total protein).Admission data collected prospectively from 148 horses with severe colic in one hospital was used to construct multivariate logistic models to predict if a horse had an infectious non-surgical colic. The models were based on 1) clinical evaluation, 2) clinical and blood evaluation and 3) clinical, blood and PF evaluation. Each model was independently validated against admission data from 78 horses in another hospital.Results and Discussion:The variables included in the final ‘clinical model’ were: Lethargy, temperature increase from 38◦C, gastric reflux 5-10L and normal rectal findings. All variables except gastric reflux 5-10L, were positive predictors of infectious colic. Diagnostic specificity and sensitivity was 98% and 57% respectively. When SAA concentration in serum was added to the model based on clinical evaluation, the specificity and sensitivity improved to 98% and 64%, respectively. No additional blood or PF variables improved the model significantly. The models had a satisfying integrity and diagnostic performance when validated.