A "risk of harm" protocol to identify youth in need of immediate emergency assistance in a study on mental health and HIV in Rwanda among 680 youth ages 10-17 is described. Cases are presented that describe the experience in using this protocol to ensure safety of participants, with ethical and logistical challenges considered. Among the population of the study, 3.2% were deemed "risk of harm." The most prevalent presenting problem was non-fatal suicidal behavior (91% of risk of harm cases), with 36% having a history of a reported previous attempt. Challenges included: acute food insecurity/significant poverty; lack of support/adequate supervision from family members; family violence; alcohol abuse; and HIV-related stigma. Development of a "risk of harm" protocol and collaboration between study staff, community leadership, health authorities, and health workers are critical to ensuring participants' safety in research among vulnerable populations.