during atria1 pacing at a rate just in excess of the sinus rate. Wellens and Durrer [I] found a strong correlation between the antegrade effective refractory period of the accessory pathway and the cycle length of the ventricular response during atria1 fibrillation. Kanakis et al. [2] found, however, that the effective refractory period depends upon the cycle length of the driving stimuli. At shorter cycle lengths the effective refractory period decreased, with a maximum decrease of 40 msec. Gallagher et al [3] confirmed this observation. The maximum decrease of the effective refractory period in their patients was 45 msec. In our case, the variability of the effective refractory period, measured at different cycle lengths, exceeded all known maximum values. Neither the effective refractory period, measured by programmed atria1 stimulation, nor the pre-Wenckebach cycle length, determined by incremental atria1 pacing, yielded values similar to the cycle length during atria1 fibrillation. Substantially shorter cycle lengths during atria1 fibrillation than the effective refractory period of the accessory pathway have not been described till now. The explanation for this phenomenon in our patient could be an unusual variability in the dependence of the effective refractory period on the driving rate. Changes in sympathetic tone could possibly cause this. We conclude that induction of atria1 fibrillation is the only reliable way of assessing those patients with Wolff-Parkinson-White syndrome who are at risk of developing fast ventricular rates during atria1 fibrillation.