OBJECTIVES: Immune checkpoint inhibitors represent important options for mUC and mNSCLC patients. Given increasing emphasis on economic value, this study examines cost differences when atezolizumab (ATEZO), nivolumab (NIVO) or pembrolizumab (PEMBRO) are used per label in these indications. METHODS: We calculated the cost of ATEZO (1200mg Q3W, 60min initial, 30min subsequent infusion), NIVO (240mg Q2W, 60min infusion), and PEMBRO (200mg Q3W, 30min infusion) based on 2017 wholesale acquisition costs (WAC) and CMS average sales price (ASP) payment rates. Dosing does not differ for the two indications. Infusion administration (ADMIN) costs were based on CMS Physician Fee Schedule ($136 for upto 60min). Since treatment durations from clinical trials have not been published, standardized costs are presented per month (30.44 days) and per 24 weeks. This analysis does not account for discounts/rebates, which is not publicly available information, and could affect the net cost to individual payers. RESULTS: Monthly costs were: ATEZO: WAC $12,449, ASP $12,427, ADMIN $198; NIVO: WAC $13,089, ASP $12,827, ADMIN $297; PEMBRO: WAC $12,895, ASP $12,733, ADMIN $198. Over 24-weeks, ATEZO patients have 4 fewer visits and 7.5 fewer infusion hours compared with NIVO; equal visits and 30min more infusion time for the first cycle compared with PEMBRO (equal after first infusion). Total (WAC+ADMIN) 24-week costs were $3,802 higher for NIVO and $2,184 higher for PEMBRO compared with ATEZO. ASP-based total 24-week costs (ASP+ADMIN) were $2,756 and $1,691 higher for NIVO and PEMBRO, respectively, compared with ATEZO. CONCLUSIONS: Immunotherapies are valuable in mUC and mNSCLC. This analysis suggests ATEZO may be less costly among cancer immunotherapies in these indications, with potential time savings for patients (fewer office visits and less infusion time vs. NIVO). Additional research is required to characterize real-world utilization and outcomes (including AEs) of immunotherapies, which might have an impact on the total cost of treatment.