Wastewater-based epidemiology uses pooled wastewater samples to monitor community health and has been used extensively during the COVID-19 pandemic to track SARS-CoV-2 RNA shed by infected individuals into wastewater. Wastewater concentrations of SARS-CoV-2 RNA have been positively correlated with contemporaneous counts of COVID-19 cases, making it useful for following relative disease burden trends within a community. However, the statistical associations are too weak for wastewater-based epidemiology to reliably predict reported case counts, limiting its potential. Here we show that wastewater SARS-CoV-2 concentrations are highly correlated with the community prevalence estimated from 8 randomized household community surveys in 6 Oregon communities over a 10-month period. We found that wastewater-based epidemiology is a significantly better predictor of COVID-19 community prevalence than reported case counts, which suffer from systematic biases including variations in access to testing and underreporting of asymptomatic cases, even after accounting for uncertainty inherent in the wastewater and prevalence estimates by using Monte Carlo simulations. Additionally, our results show that wastewater-based epidemiology can identify the rise and fall of neighborhood-scale COVID-19 hot spots and provide rapid information about the presence of SARS-CoV-2 variants at the neighborhood- and city-scale through sequence analyses of the wastewater. These results validate the potential of wastewater-based epidemiology to be a quantitative method to predict the prevalence of SARS-CoV-2 and identify the presence of variants of concern in a given community or neighborhood, independent of availability and access to individual-level testing. These advantages in combination with its scalability, relatively modest cost and low labor requirements, makes integrating permanent wastewater-based epidemiology infrastructure into public health systems a key component in creating pandemic-resilient cities in the future.