Quantitative SARS-CoV-2 antibody assays against the spike (S) protein are useful for monitoring immune response after infection or vaccination. We compared the results of three chemiluminescent immunoassays (CLIAs) (Abbott, Roche, Siemens) and a surrogate virus neutralization test (sVNT, GenScript) using 191 sequential samples from 32 COVID-19 patients. All assays detected >90% of samples collected 14 days after symptom onset (Abbott 97.4%, Roche 96.2%, Siemens 92.3%, and GenScript 96.2%), and overall agreement among the four assays was 91.1% to 96.3%. When we assessed time-course antibody levels, the Abbott and Siemens assays showed higher levels in patients with severe disease ( < 0.05). Antibody levels from the three CLIAs were correlated (r = 0.763-0.885). However, Passing-Bablok regression analysis showed significant proportional differences between assays and converting results to binding antibody units (BAU)/mL still showed substantial bias. CLIAs had good performance in predicting sVNT positivity (Area Under the Curve (AUC), 0.959-0.987), with Abbott having the highest AUC value ( < 0.05). SARS-CoV-2 S protein antibody levels as assessed by the CLIAs were not interchangeable, but showed reliable performance for predicting sVNT results. Further standardization and harmonization of immunoassays might be helpful in monitoring immune status after COVID-19 infection or vaccination.