Convalescent plasma (CP) have been used for treatment of COVID-19, but their effectiveness varies significantly. Moreover, the impact of CP treatment on the composition of SARS-CoV-2 antibodies in COVID-19 patients and antibody markers that differentiate between those who survive and those who succumb to the COVID-19 disease are not well understood. Herein, we performed longitudinal analysis of antibody profile on 115 sequential plasma samples from 16 hospitalized COVID-19 patients treated with either CP or standard of care, only half of them survived. Differential antibody kinetics was observed for antibody binding, IgM/IgG/IgA distribution, and affinity maturation in 'survived' vs. 'fatal' COVID-19 patients. Surprisingly, CP treatment did not predict survival. Strikingly, marked decline in neutralization titers was observed in the fatal patients prior to death, and convalescent plasma treatment did not reverse this trend. Furthermore, irrespective of CP treatment, higher antibody affinity to the SARS-CoV-2 prefusion spike was associated with survival outcome, while sustained elevated IgA response was associated with fatal outcome in these COVID-19 patients. These findings propose that treatment of COVID-19 patients with convalescent plasma should be carefully targeted, and effectiveness of treatment may depend on the clinical and immunological status of COVID-19 patients as well as the quality of the antibodies in the convalescent plasma.