Background Horses with non‐strangulating intestinal infarction (NSII) are often misdiagnosed with idiopathic peritonitis or acute colitis. Early diagnosis is essential to ensure early surgical intervention and improve survival. Methods Clinical and laboratory data from horses admitted to the University of Copenhagen Large Animal Teaching Hospital with NSII, idiopathic peritonitis or acute colitis between 2009 and 2018 were used for univariate comparisons and a multivariable logistic regression model for prediction of NSII. Results Two hundred and thirty‐one horses were included. A multivariable model for the prediction of NSII included gastric reflux (more than 5 L) (odds ratio [OR] 8.7; 95% confidence interval [CI] 2.1–36.2), abnormal findings palpated per rectum (intestinal dilatations/impactions [OR 4.43; 95% CI 1.43–13.38], colon displacements [OR 23.16; 95% CI 5.26–101.97] or intestinal mass [OR 179.7; 95% CI 23.5–1375.5]), white blood cell count (OR 1.2; 95% CI 1.1–1.4), packed cell volume (OR 0.9; 95% CI 0.8–0.9), age (OR 0.9; 95% CI 0.8–1.0) and heart rate (OR 1.1; 95% CI 1.0–1.1). The model had a low false positive rate (5%), but a high false negative rate (50%). Limitations Due to the retrospective nature of the study, sample collection was inconsistent, resulting in missing values. Conclusion The model had some capability in predicting NSII. However, the high risk of false negatives means that exploratory laparotomy should be considered in horses with peritonitis of unknown aetiology in areas where Strongylus vulgaris is prevalent and occurrence of idiopathic peritonitis is low.