There are limited data on how long neutralizing antibody (NAb) response elicited via primary SARS‐CoV‐2 infection will last. Eighty‐four serum samples were obtained from a prospective cohort of 42 laboratory‐confirmed COVID‐19 inpatients at the time of discharge from the hospital and in the late convalescent phase. A virus neutralization assay was performed to determine the presence and titers of NAbs with authentic SARS‐CoV‐2. Long‐term dynamics of NAbs and factors that may have an impact on humoral immunity were investigated. Mild and moderate/severe patients were compared. The mean sampling time was 11.12 ± 5.02 days (4–28) for the discharge test and 268.12 ± 11.65 days (247–296) for the follow‐up test. NAb response was present in 83.3% of the patients about 10 months after infection. The detectable long‐term NAb rate was significantly higher in mild patients when compared to moderate/severe patients (95.7% vs. 68.4%, p = 0.025). In the follow‐up, NAb‐positive and ‐negative patients were compared to determine the predictors of the presence of long‐term humoral immunity. The only significant factor was disease severity. Patients with mild infections have more chance to have NAbs for a longer time. Age, gender, and comorbidity did not affect long‐term NAb response. NAb titers decreased significantly over time, with an average rank of 24.0 versus 19.1 (p = 0.002). Multivariate generalized estimating equation analysis revealed that no parameter has an impact on the change of NAb titers over time. The majority of the late convalescent patients still had detectable low levels of neutralizing antibodies. The protective effect of these titers of NAbs from re‐infections needs further studies.