Background: The aim of this prospective, randomized, blinded study was to compare analgesic efficacy of continuous paravertebral and epidural analgesia for post-thoracotomy pain. Methods: Forty-two ASA physical status II–III patients undergoing lung resection surgery were randomly allocated to receive post-thoracotomy analgesia with either a thoracic epidural (group EPI, n = 21) or paravertebral (group PVB, n = 21) infusion of 0.2% ropivacaine (infusion rate: 5–10 mL h−1). The degree of pain at rest and during coughing, haemodynamic variables and blood gas analysis were recorded every 12 h for the first 48 h. Results: The area under the curve of the visual analogue pain score during coughing over time was 192 (60–444) cm h−1 in group EPI and 228 (72–456) cm h−1 in group PVB (P = 0.29). Rescue morphine analgesia was required in four patients of group EPI (19%) and five patients of group PVB (23%) (P = 0.99). The PaO2/FiO2 ratio reduced significantly from baseline values in both groups without between-group differences. The median (range) percentage reduction of systolic arterial pressure from baseline was −9 (0 to −9)% in group PVB and −17 (0 to −38)% in group EPI (P = 0.02); while clinically relevant hypotension (systolic arterial pressure decrease >30% of baseline) was observed in four patients of group EPI only (19%) (P = 0.04). Patient satisfaction with the analgesia technique was 8.5 (8–9.8) cm in group EPI and 9 (7.5–10) cm in group PVB (P = 0.65). Conclusions: Continuous thoracic paravertebral analgesia is as effective as epidural blockade in controlling post-thoracotomy pain, but is associated with less haemodynamic effects.