Youth are particularly vulnerable to acquiring HIV, yet reaching them with HIV prevention interventions and engaging and retaining those infected in care and treatment remains a challenge. We sought to determine the incidence rate of loss to follow-up (LTFU) and explore socio-demographic and clinical characteristics associated with LTFU among HIV-positive youth aged 15-21 years accessing outpatient care and treatment clinics in Kisumu, Kenya. Between July 2007 and September 2010, youth were enrolled into two different HIV care and treatment clinics, one youth specific and the other family oriented. An individual was defined as LTFU when absent from the HIV treatment clinic for ≥ 4 months regardless of their antiretroviral treatment status. The incidence rate of LTFU was calculated and Cox regression analysis used to identify factors associated with LTFU. A total of 924 youth (79% female) were enrolled, with a median age of 20 years (IQR 18-21). Over half, (529 (57%)), were documented as LTFU, of whom 139 (26%) were LTFU immediately after enrolment. The overall incidence rate of LTFU was 52.9 per 100 person-years (p-y). Factors associated with LTFU were pregnancy during the study period (crude HR 0.68, 95% CI 0.53-0.89); CD4 cell count >350 (adjusted hazard ratios (AHR) 0.59, 95% CI 0.39-0.90); not being on antiretroviral therapy (AHR 4.0, 95% CI 2.70-5.88); and non-disclosure of HIV infection status (AHR 1.43, 95% CI 1.10-1.89). The clinic of enrolment, age, marital status, employment status, WHO clinical disease stage and education level were not associated with LTFU. Interventions to identify and enrol youth into care earlier, support disclosure, and initiate ART earlier may improve retention of youth and need further investigation. Further research is also needed to explore the reasons for LTFU from care among HIV-infected youth and the true outcomes of these patients.