This study was undertaken to compare the effect of 1 year hemodialysis (HD) or hemodiafiltration (HDF) treatment on peripheral neuropathy. Thus 21 of 42 patients on chronic HD (1–1.3 m2 cuprophane dialyzer, Qb 300 ml/min) were switched to HDF (1.3 m2 polysulfone dialyzer, Qb 400 ml/min, substitution volume 9-13 liters, ultrafiltration rate 60-70 ml/min), while the remaining patients were considered as a control group. Treatment time was scheduled both in HD and HDF to maintain adequate BUN levels in relation to protein catabolic rate. However, HDF provided a significantly greater weekly inulin (MW 5,000) clearance than HD (5.8 ± 1.2 vs. 1.6 ± 0.2 ml/min; p < 0.001). HD and HDF groups were comparable for age, time on dialysis and starting electroneurographic parameters, which were on average within the normal range. After 1 year follow-up, creatinine, hematocrit, calcium, phosphate, PTH, BUN, protein catabolic rate and residual GFR were comparable in the two groups, whereas β2-microglobulin was significantly reduced in HDF patients (29 ± 6.7 vs. 38.8 ± 13.9 mg/l in HD patients, p < 0.01). During the 1-year treatment, electroneurographic parameters did not change in HDF patients, whereas a significant decrease of ulnar motor nerve conduction velocity, ulnar muscle action potential amplitudes, median sensory nerve conduction velocity and peroneal muscle action potential amplitudes was detected in HD patients. We conclude that HDF might prevent the worsening of the electroneurographic indices occurring during chronic HD treatment, as it provides a more effective removal of middle and larger molecules than HD. The use of a more bio-compatible membrane in HDF might further contribute to this favorable effect on uremic neuropathy.