This study investigated whether the irregularity of ventricular cycle length during atrial fibrillation (AF) is affected by digoxin. Patients (n = 41) with paroxysmal AF enrolled in a randomized crossover comparison of digoxin and placebo underwent 24-hour ambulatory monitoring during each treatment. Tapes containing AF episodes lasting at least 2 minutes were selected (24 recordings on placebo and 17 on digoxin). The mean (mRR) and standard deviation (SDRR) of RR intervals was calculated for each 30-second segment of AF. The resulting SDRR values were clustered according to bins of mRR values ranging from 350-650 ms in 25-ms steps. In each bin, the SDRR values of all placebo and all digoxin recordings were statistically compared for the top 5, 10, and 15 percentiles of each bin which represented the extremes of ventricular cycle length irregularity during AF. There were no significant differences between the total data of SDRR values in individual bins of mRR. However, the top 5, 10, and 15 percentiles of SDRR values corresponding to mRR values from 350-550 ms were significantly reduced by digoxin (P < 0.0001). The study concludes that although digoxin does not influence the mean variability of RR cycles during AF paroxysms, it suppresses episodes in which a fast ventricular response is associated with extreme variability of RR periods.