The influence of infection-control practices on bloodstream infection (BSI) risk was examined in a home health care setting in which three needleless devices were used consecutively. A case-control study and a retrospective cohort study were conducted. Risk factors for BSI included lower education level, younger age, having a central venous catheter (CVC) with multiple ports, or having a tunneled CVC. Among patients with a tunneled CVC, those at greatest risk had been allowed to shower rather than bathe and to get their exit site wet (P<.01). A high proportion (49%) of isolates were hydrophilic gram-negative bacteria, suggesting water sources of infection. In the cohort study, the BSI rate decreased as the frequency of changing the needleless device end cap increased from once weekly up to every 2 days, suggesting that the mechanism for BSI may involve contamination from the end cap. These findings may help to develop infection-control measures specific to home health care.