A prospective study was carried out to determine the clinical effect of a newly devised catheter connection method (I system) and piggyback access system. Previous studies have demonstrated that the I system avoided bacterial contamination in vitro during tubing change that Luer-Lock connectors did not. The purpose of this study was to investigate the ability of this device coupled with a new closed-system piggyback technique for multipurpose access to reduce catheter-related sepsis in clinical practice. Two hundred and thirty patients receiving total parenteral nutrition were divided into two groups. Group I (n = 106) used the I system connector and group L (n = 124) used a Luer-Lock connector. Catheters in both groups were used for multipurpose access for infusion and blood sampling. In group L, a three-way stopcock and/or pig-gyback system was used for multiple access. In group I, a newly designed closed-system piggyback was used. The incidence of catheter-related sepsis was significantly lower in group I (1.89%/catheter) than in group L (12.10%/catheter) (p < .01, chi 2 analysis), and the average duration of use of each catheter was significantly longer in group I than in Group L (p < .01 by generalized Wilcoxon test). The results of this clinical study suggest that the newly designed connection method and piggyback access system are able to reduce catheter-related sepsis.