To account for delays between specimen collection and report, the New York City Department of Health and Mental Hygiene used a time-correlated Bayesian nowcasting approach to support real-time COVID-19 situational awareness. We retrospectively evaluated nowcasting performance for case counts among residents diagnosed during March-May 2020, a period when the median reporting delay was 2 days. Nowcasts with a 2-week moving window and a negative binomial distribution had lower mean absolute error, lower relative root mean square error, and higher 95% prediction interval coverage than nowcasts conducted with a 3-week moving window or with a Poisson distribution. Nowcasts conducted toward the end of the week outperformed nowcasts performed earlier in the week, given fewer patients diagnosed on weekends and lack of day-of-week adjustments. When estimating case counts for weekdays only, metrics were similar across days the nowcasts were conducted, with Mondays having the lowest mean absolute error, of 183 cases in the context of an average daily weekday case count of 2,914. Nowcasting ensured that recent decreases in observed case counts were not overinterpreted as true declines and supported health department leadership in anticipating the magnitude and timing of hospitalizations and deaths and allocating resources geographically.