Hypoxia occurs frequently during routine hemodialysis (HD). In this study the effect of dialysate temperature on arterial blood gas parameters was investigated. Ten stable HD patients (2 smokers) were dialyzed for 240 min with each of three different dialysate temperatures: 36.5 degrees C (normal temperature HD; NHD), 38.5 degrees C (warm HD; WHD) and 34.5 degrees C (cold HD; CHD). A cuprophane plate dialyzer was used. The ultrafiltration volume was identical in each patient. Arterial blood gas samples were frequently (approximately 10 times/treatment) taken during the dialysis and immediately analyzed. The dialysate temperature significantly affected PaO2 (p < 0.001) but not PaCO2. We also compared the effect of NHD with that of WHD and CHD, respectively, as regards PaO2. NHD and WHD differed significantly p < 0.01), whereas NHD and CHD were not significantly different. However, the relative PaO2 value (% of the baseline value) at the end of CHD (105 +/- 5%) was significantly higher than after both NHD (96 +/- 4%, p < 0.01) and WHD (91 +/- 3%, p < 0.01). In the case of NHD and WHD the fraction of time during which the patients had a PaO2 value below 80 mm Hg was 62 and 64%, respectively. The corresponding figure for CHD was 44%. Arterial oxygen saturation (SaO2) increased during CHD from 95.2 +/- 0.6 to 96.7 +/- 0.6% (p < 0.05), while SaO2 was unchanged during NHD and WHD. The positive effect of CHD was evident in 7 patients. In 1 patient PaO2 was not affected by the dialysate temperature, while in the remaining 2 patients (smokers) a decrease in PaO2 was induced by WHD as well as CHD. A separate statistical analysis with the 2 smokers excluded was performed, which showed that the dialysate temperature significantly affected PaO2 (p < 0.001). A comparison between NHD and CHD showed a significant difference (p < 0.001), whereas NHD and WHD did not differ significantly. When the 2 smokers were excluded from the analysis the fraction of time with a PaO2 value below 80 mm Hg was 60% during NHD and 56% during WHD, but it was reduced to 31% during CHD. In conclusion, despite the existence of interindividual variations most patients seemed to benefit from cold dialysate for the prevention of dialysis-induced hypoxia.