Softwood kraft pulp was subjected to a laboratory one- and three-stage oxygen delignification process. Pulp and liquor samples were collected at different stages of the process with particular attention being paid to the early and late stages. A novel residual lignin isolation method extracted about 65% of the oxidized residual lignins at a purity exceeding 90%. Using this methodology coupled to quantitative 31P NMR, 13C NMR, 2D heteronuclear (HMQC) NMR spectroscopic and analytical pyrolysis techniques allowed a thorough characterization of the residual and solubilized lignin fractions at the various stages of the process. Our conclusions do not point to a single factor as being responsible for the limits confronting oxygen delignification. Amongst the major factors impeding the effectiveness of oxygen delignification are: (i) the accumulation of relatively inert 5,5' biphenyl structures originally present in kraft lignin from dibenzodioxocin ring opening reactions; and (ii) the accumulation of considerably less reactive p-hydroxyphenyl structures. Detailed quantitative information was obtained and discussed in relation to the presence and role of these unreactive phenolic fractions on the residual oxidized lignins. As such we arrived at important conclusions as to why and how these structures remain and accumulate on the fiber. In addition, issues related to the profiles of the total phenolic hydroxyl content of the residual lignins and the remaining arylglycerol β-O-4 structures are discussed.Key words: analytical pyrolysis, biphenyl structures, HMQC, NMR, oxidation mechanism, oxygen delignification, p-hydroxyphenyl moieties, quantitative 13C NMR, quantitative 31P NMR, residual lignin, solubilized lignin, total phenolic hydroxyls.