Therapeutic proteins conjugated with branched poly(ethylene glycol) (PEG) have extended in vivo circulation half-lives compared to linear PEG-proteins, thought to be due partly to a greater hydrodynamic volume of branched PEG-proteins, which reduces the glomerular sieving coefficient. In this paper, viscosity radii of PEGylated alpha-lactalbumin (M(r) = 14.2 kDa) and bovine serum albumin (M(r) = 67 kDa) prepared with linear and branched PEGs (with nominal molecular weights 5, 10, 20 and 40 kDa) were compared experimentally using size exclusion chromatography (SEC). PEG adductprotein molecular weight ratios of the PEGylated proteins covered the range 112 to 61. Direct comparisons of experimentally measured viscosity radii were found to be misleading due to differences between actual and nominal molecular weights of the PEG reagents used. Comparison with predicted viscosity radii shows that there is no significant difference between the viscosity radii of branched and linear PEG-proteins having the same total molecular weight of PEG adducts. Therefore, longer in vivo circulation half-lives of branched PEG-proteins compared to linear PEG-proteins are not explained by size difference. It is also calculated that the molecular size cut-off for glomerular filtration, 60 A for a 30 kDa PEG, matches the 30-50 A size range for the pores of the glomerular basement membrane. Finally, it is confirmed that prediction of PEG-protein viscosity radii should be based upon conservation of the total PEG adduct surface area to volume ratio for both linear and branched PEG-proteins regardless of PEGylation extent.