BACKGROUND Treatment with recombinant human erythropoietin (rHu EPO) in dialysis patients has been shown to be highly effective in terms of correcting anaemia and improving quality of life. There is debate concerning the benefits of rHu EPO use in pre-dialysis patients. There is a concern that rHu EPO may accelerate the deterioration in renal function, however the opposing view is that if rHu EPO is as effective in pre-dialysis patients that by improving the patients sense of well-being the onset of dialysis could be delayed. OBJECTIVE To assess the effects of rHu EPO use in pre-dialysis patients with renal anaemia. METHODS We searched MEDLINE (1980 to May Week 3 2001), EMBASE (1984 to Week 24 2001), BIOSIS (1985 to January 1997), CINAHL (1982 to October 1997), The Cochrane Library (Issue 1, 1997), CHEMABS (1984 to November 1996), SIGLE (1980 to June 1996), CRIB (10th edition, 1995), UK NRR (14TH consolidation, September 1996), RSC ( 1980 to February 1997), HealthSTAR (1995 to October 1997), IBSS (1984 to July 1997), NEED (July 1997) and reference lists of relevant articles. We contacted biomedical companies and investigators in the field and we hand searched Kidney International (including all supplements but excluding all conference proceedings except for 1994) July 1983 to May 1997 inclusive. The internet was also searched on: August 1997. We had also identified some studies from a previous broad search for all randomised controlled trials (RCTs) relevant to the management of end-stage renal disease. Date of the most recent search: June 2001. METHODS RCTs or quasi-RCTs comparing the use of rHu EPO with no rHu EPO or placebo in pre-dialysis patients. METHODS Only published data were used. Data were abstracted by a single investigator onto a standard form. A sample of the data abstracted was double-checked by another reviewer. The data abstracted were relevant to the predetermined outcome measures. Some authors were contacted to clarify how patients were allocated to groups. All authors from included studies were contacted for missing information. RESULTS Twelve studies with a total of 232 participants met the inclusion criteria and where possible data from these were summated by meta-analyses (Peto's Odds Ratio (OR) and Weighted Mean Difference (WMD)). The majority of the trials included small numbers and were of short duration (8-10 weeks) with the exception of three trials. There was a marked improvement in haemoglobin (mean difference 2.3g/dL, 95% CI 1.37 to 3.23) and haematocrit (WMD 9.92%, 95% CI 8.78 to 11.05) with the treatment and a decrease in the number of patients requiring blood transfusion (OR 0.25, 95% CI 0.09 to 0.69). The data from all studies which reported quality of life or exercise capacity demonstrated an improvement in the rHu EPO group. None of the measures of progression of renal disease (when a summary statistic was calculated) demonstrated a statistically significant difference. Though the requirement for antihypertensive treatment appears to be increased by rHu EPO (OR 1.84, 95% CI 1.02 to 3.32), there was no other statistically significant increase in adverse events. Based on the limited current evidence, decisions therefore have to be made on whether the putative benefits in terms of quality of life identified in the review are worth the extra costs of pre-dialysis rHu EPO. CONCLUSIONS This review has shown that treatment with rHu EPO in pre-dialysis patients corrects anaemia and avoids the requirement for blood transfusions. There are also improvements in quality of life and exercise capacity. There may be increased hypertension. Most of the trials were not of sufficient duration to assess the effects of rHu EPO on progression of renal disease. In the long term, questions still remain about whether pre-dialysis rHu EPO either speeds up or delays the onset of dialysis. Thus there is insufficient evidence on the total costs and benefits of treating pre-dialysis patients with rHu EPO.