Remote and hybrid modes of instruction have been employed as alternatives to in-person instruction as part of mitigation efforts in response to the COVID-19 pandemic. Little is known about whether instructional mode in a given school district may impact transmission in the surrounding community. Utilizing weekly case counts by zip code from August – December 2020, we employed a generalized estimating equations approach to estimate the association between instructional mode of the public school district and emergent COVID-19 cases in Cuyahoga County, Ohio. Of 51 zip codes, 37 are contained largely within a single district; of the 20 included school districts, 7 employed only remote instruction (RI) and 13 used some non-remote instruction (NRI) (2-15 weeks). Districts attempting NRI had lower average Social Vulnerability Index (SVI) (0.32 versus 0.73). Weekly incidence increased in all zip codes from August to peaks in late November to early December before declining. The zip code cumulative incidence within school districts that employed some NRI was higher than in those offering only RI (risk ratio = 1.12; risk difference = 519 per 100,000), a contrast to relatively lower pre-study incidence in NRI zip codes (incidence rate ratio of study: pre-study period = 6.12 (NRI) versus 3.59 (RI)). The mean effect (incidence rate ratio) for NRI on emergent cases 2 weeks after mode exposure, controlling for SVI was significant only for high SVI zip codes 1.30 (p<0.001). NRI may be associated with increased community COVID-19 incidence. Vulnerable communities may need more resources to open schools safely.