The primary aim of this multicenter, prospective, randomized cross-over study was to clarify whether a new model of hemodialysis (HD) potassium (K) removal using a decreasing intra-HD dialysate K concentration and a constant plasma-dialysate K gradient (treatment B) is capable of reducing the arrhythmogenic effect of standard HD, which has a constant dialysate K concentration and decreasing plasma-dialysate K gradient (treatment A). The secondary aim was to verify whether this new model is clinically safe. In treatment B, the initial dialysate K concentration had to be 1.5 mEq/liter less than the plasma K concentration, and exponentially decrease to 2.5 mEq/liter at the end of HD. Forty-two chronic HD patients with an increase in premature ventricular complexes (PVC) during dialysis were enrolled from 18 participating centers, and randomly assigned to either sequence 1 (ABA) or sequence 2 (BAB). A pool of 333 of 378 expected ECG Holter recordings were checked for signal quality; 269 (71%) from 36 patients (86%) had a satisfactory signal quality and 108 were selected for analysis (1 per patient per period). There was a difference in the natural logarithm of the increase in PVC/hr and PVC couplets/hr during HD between treatments A and B (1.70 +/- 1.59 vs. 1.09 +/- 1.76 and 0.94 +/- 0.86 vs. 0.64 +/- 1.01, a reduction of 36% and 32%, P = 0.011 and 0.047, respectively) without any carry over effect (P = 0.61 and 0.24, respectively). The fact that this decrease of one third is due to a lower plasma-dialysate K gradient is supported by the observation that it was more evident during the first than the last two hours of HD (a reduction in the natural logarithm of the increase in PVC/hr and PVC couplets/hr of 60% and 60%, P 0.002 and 0.009, vs. 26% and 17%, P = 0.098 and 0.332, respectively): the initial plasma-dialysate K gradient was 2.3 times lower during treatment B than during treatment A, without adversely affecting pre-HD plasma K levels. These results could have a considerably clinical impact not only because of the possibility of physiologically decreasing the arrhythmogenic effect of HD, but also because this effect can be considered a "marker" of the electrophysiological derangement induced by the administration of standard HD three times a week for years ("electric disequilibrium syndrome").