The binding of penetratin, a peptide that has been found useful for cellular delivery of large hydrophilic molecules, to negatively charged vesicles was investigated. The surface charge density of the vesicles was varied by mixing zwitterionic dioleoylphosphatidylcholine (DOPC) and negatively charged dioleoylphosphatidylglycerol (DOPG) at various molar ratios. The extent of membrane association was quantified from tryptophan emission spectra recorded during titration of peptide solution with liposomes. A singular value decomposition of the spectral data demonstrated unambiguously that two species, assigned as peptide free in solution and membrane-bound peptide, respectively, account for the spectral data of the titration series. Binding isotherms were then constructed by least-squares projection of the titration spectra on reference spectra of free and membrane-bound peptide. A model based on the Gouy-Chapman theory in combination with a two-state surface partition equilibrium, separating the electrostatic and the hydrophobic contributions to the binding free energy, was found to be in excellent agreement with the experimental data. Using this model, a surface partition constant of approximately 80 M(-)(1) was obtained for the nonelectrostatic contribution to the binding of penetratin irrespective of the fraction of negatively charged lipids in the membrane, indicating that the hydrophobic interactions are independent of the surface charge density. In accordance with this, circular dichroism measurements showed that the secondary structure of membrane-associated penetratin is independent of the DOPC/DOPG ratio. Experiments using vesicles with entrapped carboxyfluorescein showed that penetratin does not form membrane pores. Studies of the cationic peptide penetratin are complicated by extensive adsorption to surfaces of quartz and plastics. By modification of the quartz cell walls with the cationic polymer poly(ethylenimine), the peptide adsorption was reduced to a tolerable level. The data analysis method used for construction of the binding isotherms eliminated errors emanating from the remaining peptide adsorption, which otherwise would prevent a proper quantification of the binding.