Background Circadian variation in the incidence of acute cardiovascula
r events is well known but has not been extensively investigated in pa
roxysmal atrial fibrillation, although the significance of this arrhyt
hmia is growing in our society with the increasing number of aged peop
le. Methods and Results We detected 150 patients with paroxysmal atria
l fibrillation in a drug-free state from among 25 500 consecutive Holt
er recordings. To determine whether the onset, maintenance, and termin
ation of paroxysmal atrial fibrillation were random events, we analyze
d the total recorded duration of arrhythmia and the incidence of and n
umber of patients with the onset, maintenance, and termination of this
arrhythmia as hourly data and as hourly probabilities. A prominent ci
rcadian rhythm of the total duration of atrial fibrillation, approxima
te to 90% of which was well explained by a single cosinusoidal functio
n, was detected with a nadir around 11 AM. Because the onset of the ar
rhythmia had little or no circadian rhythm, this finding was due to a
diurnal pattern of maintenance and termination, both of which were wel
l expressed by a double-harmonic density function. Maintenance showed
a trough at 11 AM, and termination showed a peak at the same time, lea
ding to the nonuniform duration of single episodes of atrial fibrillat
ion throughout the 24-hour day. Conclusions Paroxysmal atrial fibrilla
tion showed a unique circadian variation that differed from the well-k
nown pattern for acute cardiovascular events, a point that should be k
ept in mind when antiarrhythmic therapy is evaluated. Identification o
f factors that regulate the circadian pattern of the maintenance and t
ermination of paroxysmal atrial fibrillation may lead to better chrono
therapy for preventing perpetuation of this arrhythmia.