Objective: to compare levels of loneliness between institutionalized and non-institutionalized populations. Methodology: exploratory research to measure the level of loneliness using a previously validated scale (este) in a group of non-institutionalized elders made up of 869 people over the age of 55 who attended elderly groups at one of Medellin´s neighborhoods, selected by random cluster sampling. The scale was also applied to a group of 379 institutionalized elders over the age of 55 residing at public or private gerontological centers. The sociodemographic variables analyzed were: gender, marital status, education and age. U tests by Mann-Whitney and Kruskal-Wallis, with significance of 0,05 were used for this research. Results: a higher score in family, social and marital loneliness was observed in institutionalized elders. The adaptation crisis was equal in both groups. The loneliness score of non-institutionalized adults was significantly different in the variables of marital status, education and age. Meanwhile, for the institutionalized adults, differences were only observed in the marital status variable. The total loneliness score was higher in institutionalized elders without a sentimental partner. Conclusions: the feeling of loneliness was overall higher in institutionalized elders. The marital status variable was associated with loneliness in both groups.