Abstract Analysts have regarded education transitions research as revealing late stage educational egalitarianism in the United States (e.g., Stolzenberg, 1994 ) and as sufficiently robust to guide policymakers (e.g., Hout, 2007 ). However, critics suggest parameter estimates are contaminated by selection bias (e.g., De Graaf & Ganzeboom, 1993 ), key parameter differences are unidentified (e.g., Cameron & Heckman, 1998 ), cross-transition comparisons are faulty, and the data are inappropriate. Useful modifications have been offered, (e.g., Breen and Jonsson, 2000 , Hauser and Andrew, 2006 , Lucas, 2001 ), but analysts have yet to comprehensively address the challenges critics pose. In response, we propose a neo-classical education transitions approach that uses fuller sets of data and models that explicitly address the primary threats to proper inference. Using this approach we re-assess the educational attainment process for a baby boomer cohort, a Generation X cohort, and a Generation Y cohort. All cohorts fail to replicate the waning coefficients pattern. Methodologically, the study responds to criticisms in a way that offers methods for continued cross-national comparative research. Substantively, the study undermines confidence that standard education transitions research can provide policy guidance and the claim of late stage egalitarianism in the United States educational attainment process.