Objectives: The present study analyzed whether estimates of health inequalities in old age are sensitive to the exclusion of people living in institutions and nonuse of proxy interviews. Method: Pooled data from the 2004 wave (n = 1,180, aged 69-100, primarily interviewed over the phone) and the 2011 wave (n = 931, aged 76-101, primarily interviewed face-to-face) of the Swedish Panel Study of Living Conditions of the Oldest Old were used to compare absolute and relative differences in disability and mortality between people with compulsory education and people with more than compulsory education. Results: Both absolute and relative health inequalities would have been underestimated in a survey that excluded institutionalized people and proxy-interviewed community dwellers. The same patterns were found in men and women and regardless of the mode of data collection (telephone or face-to-face interview). The degree of underestimation was lower in those 85 years and older than in those 69 to 84 years. Discussion: A survey that only includes people who live in the community and can participate without the help of a proxy might give the impression that those with low levels of education have less extensive health disadvantages than they actually have.