G-CSF (granulocyte colony-stimulating factor) treatment has been shown to cause beneficial effects including a reduction of inducible arrhythmias in rodent models of ischemic cardiomyopathy. The aim of the present study was to test whether these effects do also apply to pacing-induced non-ischemic heart failure. In 24 female rabbits, heart failure was induced by rapid ventricular pacing. 24 rabbits were sham operated. The paced rabbits developed a significant decrease of ejection fraction. 11 heart failure rabbits (CHF) and 11 sham-operated (S) rabbits served as controls, whereas 13 sham (S-G-CSF) and 13 heart failure rabbits (CHF-G-CSF) were treated with 10 μg/kg G-CSF s.c. over 17 ± 4 days. G-CSF treatment caused a ~25% increased arterial and capillary density and a ~60% increased connexin 43 expression in failing hearts. In isolated, Langendorff-perfused rabbit hearts eight monophasic action potential recordings showed prolongation of repolarization in CHF as compared with controls in the presence of the QT prolonging agent erythromycin (+33 ± 12 ms; p < 0.01). Moreover, a significant increase in dispersion of repolarization contributed to a significantly higher rate of ventricular tachyarrhythmias in CHF. G-CSF-pre-treated hearts showed a further increase in prolongation of repolarization as compared with S and CHF. The further increase in dispersion of repolarization [S-G-CSF: +23 ± 9 ms (spatial), +13 ± 7 ms (temporal); CHF-G-CSF: +38 ± 14 ms (spatial), +10 ± 4 ms (temporal); p < 0.05 as compared with S and CHF], increased the incidence of ventricular tachyarrhythmias. In summary, chronic G-CSF treatment has moderate beneficial effects on parameters potentially related to hemodynamic function in the non-ischemic rabbit CHF model. However, a significant reduction of repolarization reserve might seriously challenge its suitability as a therapeutic agent for chronic CHF therapy.