Patients newly attending the government HIV clinic in Hong Kong were studied for the prevalence and characteristics of recent HIV infection, which was defined as having a negative HIV antibody test and/or seroconversion illness within one year of a first positive antibody result. Fifty-nine (12.0%) of 492 HIV-positive patients first seen from 2001 to 2004 were determined to be recently infected. This likely represented the lower bound of the real situation. Compared with non-recent infections on univariate analysis, recent cases were more likely to be men who have sex with men (OR 2.23; 95%CI, 1.23-4.05), never married (OR 1.96; 95%CI, 1.03-3.89), had tertiary or above education (OR 3.93; 95%CI, 1.65-10.09) and with a baseline CD4>=500 cells/ul (OR 3.65; 95%CI, 1.87-6.93). Upon multivariate analysis, tertiary or above education (adjusted OR 4.23; 95%CI, 1.76-10.16) and CD4>=500 cells/ul at diagnosis (adjusted OR 3.58; 95%CI, 1.88-6.84) remained independent variables. HIV clinics are feasible settings for collecting epidemiological information of on-going infection. Differences in the profile between recent and non-recent cases may shed light on targeting efforts to prevent new HIV infections.