We hypothesized that the 'residual' thromboembolic risk in therapeutically anticoagulated patients undergoing cardioversion could potentially be related to abnormal haemorheology and platelet activation. To test this hypothesis, we firstly investigated the role of haemorheology and platelet activation in patients with paroxysmal and persistent atrial fibrillation (AF), who were compared with healthy controls and patients with permanent AF. Second, we compared these indices in patients with persistent AF, before and after successful cardioversion. We measured indices of haemorheology (haematocrit, plasma viscosity, and fibrinogen), fibrin D-dimer (an index of thrombogenesis and fibrin turnover) and platelet activation (as assessed by platelet aggregation and plasma levels of beta-thromboglobulin, and soluble P-selectin) in 29 patients with paroxysmal AF, 87 patients with permanent AF and 29 healthy controls in sinus rhythm. The effects of cardioversion were studied in 20 patients with persistent AF, who maintained sinus rhythm at 2 months follow-up. Plasma levels of beta-thromboglobulin (P = 0.03) and fibrin D-dimer (P = 0.001) were higher in patients with AF, when compared with controls; the highest levels were seen in those with permanent AF (Tukey's test, < 0.05). Plasma viscosity was significantly higher in the patients with paroxysmal AF compared with healthy controls (P = 0.02). Plasma soluble P-selectin levels and platelet aggregation responses to all four platelet agonists (adenosine diphosphate, collagen, epinephrine and thrombin) in patients with paroxysmal AF and permanent AF were similar to controls. Plasma fibrinogen, viscosity and other markers of platelet activation (including platelet aggregation) were not significantly different in patients with paroxysmal AF, during episodes of AF and sinus rhythm (P = not significant), although mean haematocrit was significantly higher during the episodes of AF compared with episodes of sinus rhythm (P = 0.03). Among the patients with persistent AF who remained in sinus rhythm at 2 months following successful cardioversion, there was a significant decrease in the plasma levels of soluble P-selectin at 2 weeks and 2 months, when compared with baseline (pre-cardioversion) levels (P < 0.001). Haemorheology and platelet aggregation response to agonists did not change significantly, except for a transient increase in platelet aggregation response to collagen at 2 weeks (P = 0.045). In conclusion. abnormal haemostatic and platelet activation in patients with permanent AF are not consistently observed in patients with paroxysmal and persistent AF. Abnormal haemorheology appears to play an important role in patients with paroxysmal AF, especially during the paroxysms of AF. Cardioversion of persistent AF to sinus rhythm appears to decrease the platelet activation, but whether this translates into a beneficial reduction in thromboembolic risk requires further study.