OBJECTIVES: Three filters were tested for in situ efficacy in reducing bacterial contamination associated with injection drug use. METHODS: In a self-matched control design with blinded laboratory testing, injection drug users were asked to use 3 filters in random succession when loading their syringes with drug solute. RESULTS: The 0.22-micron filter proved significantly better than both the cigarette filter (relative risk [RR] = 18.0) and the 20-micron filter (RR = 4.5) in rendering syringes bacteria-free. CONCLUSIONS: The 15- to 20-micron syringe filter currently provided injection drug users in Switzerland does not significantly reduce contamination associated with common bacterial infections among users. Filters with pore width 1/100th as large are recommended.