Non-return valves (NRVs) are designed to avoid backflow of infusion fluid against the designated direction of flow (DDF) when more than one infusion is delivered via one venous access. We tested in vitro whether NRVs reliably prevent flow against the DDF at clinically relevant low flow rates. Since catheter-related infections caused by the infusion of contaminated fluids represent a relevant problem in patient care, we tested whether NRVs preclude bacterial contamination of infusions proximal to the NRVs and thus might play a role in preventing healthcare-associated infections. Additionally, the incidence of bacterial contamination of drips and infusion fluids in our intensive care unit (ICU) was quantified. In vitro, a low flow against the DDF of ten examples each of five different NRV models was applied and the integrity for fluid and transmigration of three different indicator micro-organisms was tested. Second, we investigated whether contamination of intravenous infusion tubing collected from patients treated on our ICU occurred. Largely independent from the model, 40% of the tested NRVs were not leak-tight for fluids when a pressure against the DDF was built up slowly. In 30%, bacteria migrated against the DDF and were detected proximal to the valve. In 6.7% of the tubing samples collected from ICU patients we detected bacterial contamination. In conclusion, contamination of drips is a relevant problem on ICU. NRVs neither reliably prevent backflow of fluids nor serve as micro-organism filters. Therefore they cannot be recommended as a way of reducing healthcare-associated infections.