Objectives: To identify the effect of health deterioration on residential dependency., Material and Methods: We performed a cross-sectional analysis of the microdata from the Catalan Survey of Health (2006), which features a sample of 3566 individuals aged 65 and over. A set of socio-demographic (sex, age, marital status, educational level and municipality size), as well as health variables (self-rated health, BADL and IADL dependency) associated with residential dependency are analysed by bivariate and multivariate logistic regression., Results: Multivariate analysis shows that age, marital status and health are the variables that most affect living arrangements and cohabitation. Among men, being aged 80 or over (OR>4), being unmarried or widowed (OR=6.4) and having one or more IADL dependencies (OR>2.8) increases the risk of residential dependency. Whereas for women being aged 80 and over (OR>4), being unmarried (OR=6.8) or widowed (OR=11.8) and having three or more IADL dependencies (OR=2.7) is associated with residential dependency. Municipality size and the level of education (in the latter case only for men) are also significant determining factors (P<0.05)., Conclusion: Although health deterioration, and especially IADL dependency, affects residential dependency, its impact is lower than that of socio-demographic variables, such as marital status or age. What is more, health has a greater influence on men than women, who live independently until they experience great difficulty in coping with their activities of daily living. On the other hand, men seem to fall more easily into residential dependency once they experience any IADL dependency., (Copyright © 2009 SEGG. Published by Elsevier Espana. All rights reserved.)