Background: Postoperative pain after total knee arthroplasty (TKA) is a continuing problem despite optimised multimodal analgesia. Previous studies have shown preoperative glucocorticoids to reduce postoperative pain, but knowledge about specific doses and effects in specific patient groups is lacking. Methods: A two-centre, double-blind, two-arm study comparing preoperative dexamethasone (1 mg kg⁻¹ vs 0.3 mg kg⁻¹ i.v.) on postoperative pain in 160 planned TKA subjects with low preoperative pain catastrophising and no opioid use. Subjects received multimodal analgesia with paracetamol, cyclooxygenase-2 inhibitors, local anaesthetic infiltration analgesia, and rescue opioids. The primary outcome was percentage of subjects experiencing moderate to severe pain (visual analogue scale >30 mm) upon ambulation at 24 h. Secondary outcomes included pain scores, postoperative inflammation (C-reactive protein), opioid and antiemetics use, and ‘Quality of Recovery-15’ and ‘Opioid-Related Symptom Distress Scale’, length of stay, readmissions, and complications up to Day 90. Results: A total of 157 subjects (80 vs 77) were included. No difference was found between groups in the incidence of subjects experiencing visual analogue scale >30 on ambulation 24 h after surgery (56% vs 53%, relative risk =1.07, confidence interval: 0.8–1.4, P=0.65). No differences in other pain outcomes or use of rescue opioids and antiemetics, in Quality of Recovery-15 and Opioid-Related Symptom Distress Scale, length of stay, readmissions, or complications. C-reactive protein values were comparable at 24 h (13 [6–25] mg L⁻¹ vs 16 [9–38] mg L⁻¹, P = 0.07), but lower at 48 h (26 [9–52] mg L⁻¹ vs 50 [30–72] mg L⁻¹, P<0.01) in the high-dose group. Conclusion: Use of 1 mg kg⁻¹ vs 0.3 mg kg⁻¹ i.v. dexamethasone in low pain responders after TKA did not improve early postoperative pain or other outcomes in contrast to benefits in a high pain responder population. Clinical trial registration: NCT03758170 (first registration 29-11-2018).