Background Single-site or multisite atrial pacing may reduce the incid
ence of atrial fibrillation in humans. The therapeutic mechanisms may
include synchronization of atrial repolarization (repolarization ''mem
ory'') and/or decreased dispersion of atrial repolarization. These res
ponses have not been well documented in intact atria. Methods and Resu
lts Monophasic action potential recordings were made from six atrial e
picardial sites in 39 isolated perfused rabbit heart preparations duri
ng 3 hours of continuous right atrial, left atrial, or biatrial pacing
. Action potential recordings obtained at times 0, 45, 90, 135, and 18
0 minutes were computer analyzed for activation time (AT) and 90% acti
on potential duration (APD) at each site. No consistent relationship c
ould be demonstrated between APD and AT at any time during atrial paci
ng (all P>.05). On average, left atrial APDs were longer than right at
rial APDs by up to 6.3 ms at all times, regardless of the site of paci
ng (P less than or equal to.05). At all times, dispersion of atrial re
polarization was minimized by left atrial pacing compared with right a
trial pacing (21.6+/-9.1 versus 32.4+/-15.1 ms, respectively, at time
0; P<.05). Biatrial pacing provided no further reduction in dispersion
of repolarization compared with left atrial pacing (all P>.05). Concl
usions No relationship can be demonstrated between atrial AT and APD i
n the isolated rabbit heart preparation. This differs from ventricular
repolarization ''memory,'' which is demonstrable under the same condi
tions. Left atrial APD is, on average, longer than right atrial APD, s
uggesting spatial heterogeneity in repolarization. Dispersion of atria
l repolarization is minimized by left atrial pacing in this preparatio
n with no further advantage to biatrial pacing.