There is a controversy as to whether the number of lumens in the central venous catheters may impact the incidence of catheter-related bloodstream infection. We performed a systematic search (MEDLINE, PREMEDLINE, Cochrane Library, EMBASE, BIOSIS Previews, CINAHL, HealthSTAR/Ovid healthstar, bibliographies, any language, to April, 2003) for full reports on randomized comparisons of single-lumen and multi-lumen catheters. Trials had to report on dichotomous data of catheter colonization or bloodstream infection. Meta-analyses were performed using a fixed effect model. Data were expressed as odds ratio (OR) and number-needed-to-treat (NNT) with 95% confidence interval (CI). Five randomized trials (1987-1995) with data on 255 single-lumen and 275 multi-lumen catheters were analyzed. Average insertion times were 8 to 21 days with multi-lumen catheters and 9 to 24 days with single-lumen catheters. In 4 trials, 23 of 176 (13.1%) multi-lumen and 26 of 177 (14.7%) single-lumen catheters were colonized (OR, 0.92; 95% CI, 0.49-1.72). In 5 trials, bloodstream infection occurred with 23 of 275 (8.4%) multi-lumen and with 8 of 255 (3.1%) single-lumen catheters (OR, 2.58; 95% CI, 1.24-5.37; NNT, 19; 95% CI, 11-75). For every 20 single-lumen catheters inserted, one bloodstream infection will be avoided that would have occurred had multi-lumen catheters been used. The risk of catheter colonization is not decreased. Although these conclusions are based on limited data, single-lumen catheters should be used whenever feasible.