OBJECTIVE--To investigate the effects of low dose sotalol on the signal averaged surface P wave in patients with paroxysmal atrial fibrillation. DESIGN--A longitudinal within patient crossover study. SETTING--Cardiac departments of a regional cardiothoracic centre and a district general hospital. PATIENTS--Sixteen patients with documented paroxysmal atrial fibrillation. The median (range) age of the patients was 65.5 (36-70) years; 11 were men. MAIN OUTCOME MEASURES--Analysis of the signal averaged P wave recorded from patients not receiving antiarrhythmic medication and after 4-6 weeks' treatment with sotalol. P wave limits were defined automatically by a computer algorithm. Filtered P wave duration and energies contained in frequency bands from 20, 30, 40, 60, and 80 to 150 Hz of the P wave spectrum expressed as absolute values (P20, P30, etc) and as ratios of high to low frequency energy (PR20, PR30, etc) were measured. RESULTS--No difference in P wave duration was observed between the groups studied (mean (SEM) 149 (4) without medication and 152 (3) ms with sotalol). Significant decreases in high frequency P wave energy (for example P60: 4.3 (0.4) v 3.3 (0.3) microV2.s, P = 0.003) and energy ratio (PR60: 5.6 (0.5) v 4.7 (0.6), P = 0.03) were observed during sotalol treatment. These changes were independent of heart rate. CONCLUSIONS--Treatment with low dose sotalol reduces high frequency P wave energy but does not change P wave duration. These results are consistent with the class III effect of the drug and suggest that signal averaging of the surface P wave may be a useful non-invasive measure of drug induced changes in atrial electrophysiology.