The objective of this study is to determine if pediatric advance care planning (pACP) increases adolescent/family congruence in end-of-life (EOL) treatment preferences longitudinally. Adolescents aged 14–21 years with HIV/AIDS and their families were randomized (N = 105 dyads) to three-60-minute sessions scheduled one week apart: either the pACP intervention (survey administered independently, facilitated conversation with adolescent and family present, completion of legal advance directive document with adolescent and family present) or an active control (developmental history, safety tips, nutrition and exercise education). This longitudinal, single-blinded, multi-site, randomized controlled trial was conducted in six pediatric hospital-based HIV-clinics, located in high HIV mortality cities. The Statement of Treatment Preferences measured adolescent/family congruence in EOL treatment preferences at immediately following the facilitated pACP conversation (Session 2), and at 3-month post-intervention. The mean age of adolescent participants was 18 years (range 14–21 years); 54% were male; and 93% were African-American. One-third had an AIDS diagnosis. Immediately post-intervention the Prevalence Adjusted Bias Adjusted Kappa showed substantial treatment preference agreement for pACP dyads compared to controls (High burden/low chance of survival, PABAK = 0.688 vs. 0.335; Functional impairment, PABAK = 0.687 vs. PABAK= 0.34; Mental impairment, PABKA = 0.717 vs. 0.341). Agreement to limit treatments was greater among intervention dyads than controls (High burden: 14.6% vs. 0%; Functional impairment = 22.9% vs. 4.4%; and Mental impairment: 12.5% vs. 4.4%). Overall treatment preference agreement among pACP dyads was high immediately post-intervention, but decreased over time. In contrast, treatment agreement among control dyads was low and remained low over time. As goals of care change over time with real experiences, additional pACP conversations are needed.