The purpose of this research was to study attachment, social support and loneliness in older adults. Tasks were to discover if (a) adult attachment in later life would replicate patterns found in prior research, (b) a stagewise block regression model including attachment would predict loneliness, (c) attachment would moderate the relationship between enacted support and loneliness, and (d) attachment would moderate the relationship between satisfaction with enacted support and loneliness.;Measures used were the UCLA Loneliness Scale-Version 3 (Loneliness), The Relationship Questionnaire (Attachment), The Inventory of Socially Supportive Behaviors (Enacted Received Social Support), and a one-item measure created for this study (Satisfaction with enacted received social support). Exogenous demographic variables were also included in the model.;The Dillman Tailored Design method was used and a mail survey (N = 1,000) was sent to a randomly selected sample of men and women, aged 65 and older, residing in a metropolitan area of the non-industrial American Midwest. Response rate was 42% (N = 378).;Results showed mean differences in attachment style, but differences were not consistent with prior findings possibly due to measurement issues. Stagewise block regression (block order determined theoretically and conceptually before entering) revealed that attachment significantly contributed to a model predicting loneliness (16% R2 change). The model explained 44% of the variance in loneliness. Secure and Preoccupied attachment each moderated the relationship of the directive guidance category of enacted social support and loneliness. Attachment did not moderate the relationship of satisfaction with enacted support and loneliness.