This study was designed to investigate the effect of intravenous (i.v.) iron dextran (ID) on hematocrit (Her), transferrin saturation (TS), and serum ferritin (SF) in hemodialysis patients treated with a constant dose of erythropoietin (EPO). The sensitivity, specificity, and predictive values of SF and TS for monitoring ID therapy were also assessed. All hemodialysis patients with baseline SF < 100 ng/mL or TS < 20%, with EPO dose unchanged 6 weeks before and 4 weeks after dosing with ID were included. ID (500 mg-l g) was given as an infusion over I h. Patients receiving packed RBC or with active bleeding were excluded. Hct,;TS, and SF were measured 2 weeks before and 4 weeks after ID. Linear correlation coefficients between dose of ID, changes in Hcc, TS, and SF were calculated. The sensitivity, specificity and predictive values of TS and SF were compared. A positive Her response was defined as a >5% increase from baseline 4 weeks after administration of ID. Thirty-three patients (17 females) received a total of 51 doses of ID. Mean +/- SD ID dose was 770 +/- 278 mg. Hct increased by a mean +/- SD of 4.8% +/- 9.9% (33.4% +/- 3.0% to 34.9% +/- 4.1% [p = 0.028]); SF rose by a median of 208.65% (mean +/- SD of 126.8 +/- 132.1 ng/mL to 325.3 +/- 222.0 ng/mL [p < 0.0001]; TS increased by a median of 53.8% (19.4% +/- 9.4% to 29.3% +/- 11.3% [p < 0.0001])SI-om baseline values. The correlations between dose of ID, and percent changes in Sf;: TS, and Net were poor (r(2) < 0.02). The sensitivities and specificities were 74% and 36% (TS < 20% alone); 60% and 30% (SF < 100 ng/mL alone); and 33% and 67% (TS < 20% and SF < 100 ng/mL), respectively. The predictive values for positive responses were 48% for TS and 45% for SF when used alone, and 47% when both indices were used together The predictive value increased to 65% when either SF < 100 ng/mL or TS < 20% were used. At a constant EPO dose, there was a statistically significant increase in Hct 4 weeks after ID administration in patients who were diagnosed with iron deficiency by using TS < 20% or SF < 100 ng/mL. The dose of ID administered was poorly correlated to changes in Her, TS, and SE;: Both TS and SF are non-specific and insensitive indicators for accurate diagnosis of iron deficiency in hemodialysis patients in EPO.