BACKGROUND The risk of spinal haematoma in patients receiving epidural catheters is estimated using routine coagulation tests, but guidelines are inconsistent in their recommendations on what to do when results indicate slight hypocoagulation. Postoperative patients are prone to thrombosis, and thromboelastometry has previously shown hypercoagulation in this setting. We aimed to better understand perioperative haemostasis by comparing results from routine and advanced tests, hypothesizing that patients undergoing major upper gastrointestinal surgery would be deficient in vitamin K-dependent coagulation factors because of malnutrition, or hypocoagulative because of accumulation of low molecular weight heparin (LMWH). METHODS Thirty-eight patients receiving epidural analgesia for major upper gastrointestinal surgery were included. We took blood at the time of preoperative epidural catheterization and at catheter withdrawal. Prothrombin time-international normalized ratio (PT-INR), activated partial thromboplastin time (aPTT) and platelet count (Plc) were analysed, and also albumin, proteins induced by vitamin K absence (PIVKA-II), rotational thromboelastometry (ROTEM®), multiple electrode aggregometry (Multiplate®) and activities of factors II, VII, IX, X, XI, XII and XIII. RESULTS Postoperative coagulation was characterized by thrombocytosis and hyperfibrinogenaemia. Mean PT-INR increased significantly from 1.0 ± 0.1 to 1.2 ± 0.2 and mean aPTT increased significantly from 27 ± 3 to 30 ± 4 s. Activity of vitamin K-dependent factors did not decrease significantly: FIX and FX activity increased. FXII and FXIII decreased significantly. Mean Plc increased from 213 ± 153 × 106/L while all mean ROTEM-MCFs (maximal clot firmnesses) especially FIBTEM-MCF increased significantly to above the reference interval. All mean ROTEM® clotting times were within their reference intervals both before and after surgery. ROTEM® (HEPTEM minus INTEM) results were spread around 0. There were significant correlations between routine tests and the expected coagulation factors, but not any of the viscoelastic parameters or PIVKA-II. Multiplate® area under curve and EXTEM-MCF correlated significantly to Plc as did EXTEM-MCF to fibrinogen, FIX, FX and FXIII; and FIBTEM-MCF to Plc, FII, FXI and FXIII. CONCLUSIONS The increase in PT-INR may be caused by decreased postoperative FVII while the elevated aPTT may be caused by low FXII. The mild postoperative hypocoagulation indicated by routine tests is not consistent with thromboelastometry. The relevance of ROTEM® and Multiplate® in the context of moderately increased routine tests remains unclear. Trial registration number is not applicable since this is not a clinical trial.