We have investigated the role of diameter, charge, and steric shielding on the brain distribution of liposomes infused by convection enhanced delivery (CED) using both radiolabeled and fluorescent-labeled particles. Liposomes of 40 and 80-nm diameter traveled the same distance but penetrated significantly less than a 10-kDa dextran; whereas 200-nm-diameter liposomes penetrated less than 80 nm liposomes. A neutral liposome shielded by polyethylene glycol (PEG; 2 kDa; 10% by mole) penetrated significantly farther than an unshielded liposome. Even when shielded with PEG, positive surface charge (10% by mole) significantly reduced the penetration radius compared to a neutral or negative charged liposome (10% by mole). A mathematical CED model including a term for liposome cell binding was applied to analyze the radius of particle penetration. Neutral liposomes had a binding constant of k=0.0010+/-0.0002 min-1, whereas for positive charged liposomes k increased 50-fold. The binding constant was independently verified using a degradable lipid radiolabel that eliminated from the brain with a 9.9+/-2.0 h half-life, equivalent to the calculated elimination constant k=0.0012+/-0.0002 min-1. During CED, liposomes accumulated in a subpopulation of perivascular cells within the brain. A non-degradable lipid radiolabel showed that lipid components remained within these perivascular brain cells for at least 2 days. To reduce this uptake, 100-fold molar excess of non-labeled liposomes were co-infused with labeled liposomes, which significantly increased liposome penetration. These studies suggest that optimization of therapeutic CED using particles such as drug-loaded liposomes, polymeric nanoparticles, non-viral DNA complexes, and viruses will require a strategy to overcome particle binding and clearance by cells within the CNS.