OBJECTIVES: In retrospective claims-based studies of drug exposures and resultant safety/tolerability issues (TIs), researchers must define an a-priori “at-risk window” – the time period in which observed TIs are classified as resulting from drug exposure. In HIV, interruption of combination antiretroviral therapy (cART) may cause detrimental clinical outcomes; thus, if a TI necessitates a switch in cART, the claims-based evidence of healthcare utilization for TIs may persist even after switching. This study quantified the impact of at-risk window assignment on retrospectively-estimated incidence and costs of TIs. METHODS: Data were Medicaid claims from 15 states. Subjects were HIV patients 18–64 years old initiating atazanavir, darunavir, fosamprenavir, or lopinavir-based cART regimens from January 1, 2003 to July 1, 2010. Outcomes: incidence and costs of new-onset medically-attended (ICD-9-CM-coded or treated) TIs (gastrointestinal, lipid abnormalities, diabetes/hyperglycemia, rash, jaundice) during a period of up-to 6 months post-cART-initiation or censoring at switch. RESULTS: Sample included 20,024 patients; mean age 42 years, 34% female, 48% black. When extending the at-risk window for 7 days post-switch, differences in incidence rates of TIs per 1,000 patient-months were: gastrointestinal (76.1 with extension vs. 73.7 without); lipid abnormalities (23.5 vs. 22.3); diabetes/hyperglycemia (9.0 vs. 8.8); rash (98.5 vs. 96.0); jaundice (0.59 vs. 0.57). When extending the cost-accrual window to include costs of TIs that persist after switching, differences in the estimated per-patient per-month (PPPM) costs of TIs were: gastrointestinal ($72 with extension vs. $75 without); lipid abnormalities ($27 vs. $22); diabetes/hyperglycemia ($71 vs. $67); rash ($7 vs. $8); jaundice ($0.4 vs. $2). CONCLUSIONS: Extension of at-risk windows increased estimated incidence rates of TIs, decreased estimated PPPM costs of acute TIs (i.e., gastrointestinal, rash, jaundice) and increased the estimated PPPM cost of chronic TIs (i.e., lipid abnormalities, diabetes/hyperglycemia). These novel results may inform future retrospective studies of TIs in cART.