Methods to infuse drugs into the parenchyma of the central nervous system (CNS) have been reported as inconsistent or unpredictable. The source of variability appears to be a compromised seal between the tissue and the outer surface of the cannula. Failure of the tissue to seal to the cannula creates a path of least resistance. Rather than penetrate the target area, the drug backflows along the path of the cannula. This artifact can be difficult to detect because drugs enter the systemic circulation and provide some fraction of the intended therapy. Decreasing the rate of the infusion can reduce backflow. However, this may not be an attractive option for certain therapeutic targets because decreased infusion rates decrease the volume of drug distribution in normal tissue. Cannula design plays a role. Rigid catheters that are fixed to the skull will oppose movements of the brain and break the seal between the catheter and the tissue during chronic infusions. Flexible infusion cannulas, which can be readily made by modifying commercially available brain infusion catheters with plastic tubing, appear to provide consistent infusion results because they can move with the brain and maintain their tissue seal.