We investigated the long-term effects of maintaining high hemoglobin (Hb) on renal function in patients with chronic kidney disease not on dialysis. Subjects (Hb < 10 g/dL and serum creatinine (Cr) 2–6 mg/dL) were randomized to either a high Hb group (N = 161, 11.0 ≤ Hb < 13.0 g/dL) receiving darbepoetin alfa or to a low Hb group (N = 160, 9.0 ≤ Hb < 11.0 g/dL) with epoetin alfa, stratified according to baseline Hb and serum Cr levels, comorbidity of diabetes, and study centers. Primary endpoints were composites of the following events: doubling of serum Cr, initiation of dialysis, renal transplantation, or death. Three-year cumulative renal survival rates (95% CI) were 39.9% (30.7–49.1%) and 32.4% (24.0–40.8%) in the high and low Hb groups, respectively (log-rank test; P = 0.111). A Cox proportional-hazards model adjusted by age, sex and the randomization factors showed a significantly lower event rate in the high Hb group (P = 0.035). The estimated hazard ratio (95% CI) for the high versus the low Hb group was 0.71 (0.52–0.98), the risk reduction was 29% in the high Hb group. Incidences of serious adverse cardiovascular events did not differ significantly between the high and low Hb groups (3.1% and 4.4%, respectively). No safety issues were noted in either group. Maintaining higher Hb levels with darbepoetin alfa better preserved renal function in patients with chronic kidney disease not on dialysis.