In contemporary society, online Social Networking Sites (SNS) such as Facebook provide increasingly popular contexts within which late adolescent peer interactions and accompanying identity experiments can occur. Consequently, of increasing interest is exploring the impact of SNS use on psychological functioning in this age group. There is some evidence suggestive of a relationship between greater SNS use and increased depressive symptoms. However, findings are inconsistent, with a large body of literature also indicative of possible beneficial effects of SNS use on adolescent social and emotional adjustment. Therefore, as a means to address this divergence, the present study aims to investigate whether it is the quantity of use, including use of the site to connect with existing or new contacts, or the quality of Facebook interactions that might relate to depressive symptoms. Moreover, the present research attempts to identify for which late adolescents these associations are more likely to be a risk, drawing on constructs implicated in offline self-definition and identity development. One hundred and sixty-nine late adolescents (mean age 18.6 years) participated in this quantitative, cross-sectional study. Participants completed an online survey comprising self-report questionnaires validated by previous research assessing depressive symptoms, the intensity of Facebook use, strategy used to connect with peers on Facebook, self-reported quality of interactions on Facebook, and self-definition and identity variables; self-concept clarity (SCC), separation-individuation, and ego-identity commitment. Consistent with previous research, no relationship was found between the intensity of Facebook use, including number of Facebook friends, time spent on the site each day, perceived integration of the platform into daily life, and connection strategy and depressive symptoms. There was, however, evidence suggestive of a relationship between reports of feeling down following interactions on Facebook and increased depressive symptoms. Self-definition and identity variables were not found to moderate this relationship.