Purpose: To compare the modeled costs of the Protocol AC bevacizumab-first treatment protocol (with a switch to aflibercept for sub-responders) with real-world costs for treatment-naïve patients with diabetic macular edema (DME) over a 2-year period. Methods: Published data from the Diabetic Retinopathy Clinical Research Network (DRCR) Protocol AC bevacizumab-first arm (154 eyes) were used to model 2-year treatment costs. Real-world costs were modeled using data from the Vestrum Health electronic medical records database from a 2016 to 2018 cohort of treatment-naïve eyes with DME (n = 1062) treated with antivascular endothelial growth factor monotherapy. The visual acuity (VA) before treatment in the real-world cohort was matched to Protocol AC. A secondary cost analysis further matched VA gains after treatment in the real-world cohort (n = 346) to Protocol AC. Results: In Protocol AC, the modeled 2-year DME treatment cost in the bevacizumab-first arm was $18,952, with a mean of 16.1 injections over 22.5 visits and 70% of eyes being switched to aflibercept by year 2. (Within 2 years, 57% of injections were bevacizumab and 43% were aflibercept.) Over the same period, the modeled 2-year real-world cost ($11,459) was 40% lower, with a mean of 8.6 injections over 13.8 visits (42% bevacizumab, 45% aflibercept, 13% ranibizumab). Even when matched for baseline VA and a 14-letter VA gain over 2 years, the real-world cost ($15,394) was still 19% lower than the modeled cost in the Protocol AC bevacizumab-first arm. Conclusions: The real-world cost of treating DME over 2 years was significantly lower than the Protocol AC bevacizumab-first cost. Application of Protocol AC findings into real-world utilization, as with step therapy mandates, should only be considered if the same intensive protocol could be followed. These data suggest that existing real-world costs that reflect physician choice are already significantly lower than protocol-mandated step therapy, even when controlling for similar VA outcomes.