The minimal necessary dose of Innohep (IH) (MNDI) (Innohep [tinzaparin], Leo Pharmaceutical Corp., Ballerup, Denmark) was examined in 40 patients switched from conventional heparin ([CH], Leo Pharmaceutical Corp.) to IH and in 13 patients already treated with IH. Clotting in the venous chamber and in the dialyzer was evaluated on a 4 point scale by visual inspection. IH was administrated as a bolus injection into the arterial side of the dialyzer at the beginning of dialysis sessions. The initial dose of IH was 50% of the total dose of CH used before the study (in respective IU). According to clotting in the venous chamber or dialyzer, the dose of IH was titrated by stepwise changes of 500 IU to the lowest possible dose until 3 subsequent dialysis sessions without clotting were obtained. The total dose of CH (bolus and infusion) before switching was 6,162 +/- 2,100 IU. The bleeding time from the cannulation site after dialysis, in 24 patients with A-V fistulas, was 7.1 +/- 2.8 min(triplicates). Eight patients were excluded before achieving the MNDI, 3 due to bleeding not clearly related to heparinization (1 due to gingival bleeding, 1 to epistaxis, and 1 to sugillations), 1 due to alopecia, 2 due to a need of more than 10,000 IU of IH, and 2 patients due to cessation of treatment resulting from anxiety. After switching over, the MNDI amounted to 66 +/- 26% in respective IU. The conversion IH/CH ratio correlated significantly to the blood flow rate and the type of dialyzer. When compared on 3 subsequent sessions before and after switching to IH, no differences were found in the bleeding time after decannulation and in clotting in the venous chamber while dialyzer clotting fell on the visual scale from an average of 0.36 to 0.19 (p < 0.01). No total clot formation was observed during the study. The MNDI correlated positively to the body weight, blood flow rate, and time on dialysis (with the respective coefficients of correlation of r being 0.58, 0.44, and 0.30, p < 0.05) and was also influenced by the type of dialyzer. The average MNDIs for the Hemoflow-FS hollow-fiber (Fresenius, Bad Homburg, Germany), Lundia PRO plate (Gambro, Lund, Sweden), and Polyflux hollow fiber (Gambro) were 2,571, 3,727, and 5,020 IU (p < 0.01, ANOVA). In patients on chronic hemodialysis, IH given as a bolus of 4,250 IU effectively prevented extracorporeal clotting during dialysis, similarly to CH. However, a considerable individual variation in MNDIs not related to the need for CH was observed, and this necessitates individual dosage adjustments to obtain the optimal prevention of clotting with minimal bleeding risk.