Growing life expectancy and changes in financial, marriage and labour markets have placed the income position of the elderly at the center of scientific and political discourse. As a consequence, the last decades witnessed the publication of various influential reports that contained comparative statistics on old age income inequalities on the basis of international surveys. Common to these surveys is that they exclude the elderly who live in institutions. The divergence between the target population (e.g. the population aged 65 and over) and the survey population (e.g. the noninstitutionalized population aged 65 and over) that thus arises, might lead to important bias in the survey results. However, hardly any research has been conducted quantifying the direction and strength of this bias. This article tries to fill this gap and assesses the consequences of excluding the institutionalized elderly for the validity and international comparability of a number of indicators. Analyses with the Belgian Datawarehouse Labour Market and Social Protection show that the resulting bias is negligible for average equivalent pension income, but that assistance dependency among pensioners is underestimated by 10%. Furthermore, on the basis of international statistics, it is shown that the share of elderly in institutions varies substantially across countries. It is argued how this jeopardizes the international comparability of old age statistics. Finally, the article opens up a discussion on the meaning of lack of income and wealth among institutionalized elderly. It concludes that depending on how this question is answered, poverty will be under- or overestimated in countries with a high share of institutionalized elderly.