Background: Treatment of COVID-19 patients with convalescent plasma containing neutralising antibody to SARS-CoV-2 is under investigation as a means of reducing viral loads, ameliorating disease outcomes, and reducing mortality. However, its efficacy might be reduced in those infected with the emerging B.1.1.7 SARS-CoV-2 variant. Here, we report the diverse virological characteristics of UK patients enrolled in the Immunoglobulin Domain of the REMAP-CAP randomised controlled trial. Methods: SARS-CoV-2 viral RNA was detected and quantified by real-time PCR in nasopharyngeal swabs obtained from study subjects within 48 hours of admission to intensive care unit. Antibody status was determined by spike-protein ELISA. B.1.1.7 strain was differentiated from other SARS-CoV-2 strains by two novel typing methods detecting the B.1.1.7-associated D1118H mutation with allele-specific probes and by restriction site polymorphism (SfcI). Findings: Of 1260 subjects, 90% were PCR-positive with viral loads in nasopharyngeal swabs ranging from 72 international units [IUs]/ml to 1.7×10 ¹¹ IU/ml. Median viral loads were 45-fold higher in those who were seronegative for IgG antibodies (n=314; 28%) compared to seropositives (n=804; 72%), reflecting in part the latter group’s possible later disease stage on enrolment. Frequencies of B.1.1.7 infection increased from early November (60%). Anti-SARS-CoV-2 seronegative individuals infected with wild-type SARS-CoV-2 had significantly higher viral loads than seropositives (medians of 1.2×10 ⁶ and 3.4 ×10 ⁴ IU/ml respectively; p=2×10 ⁻⁹ ). However, viral load distributions were elevated in both seropositive and seronegative subjects infected with B.1.1.7 (13.4×10 ⁶ and 7.6×10 ⁶ IU/ml; p=0.18). Interpretation: High viral loads in seropositive B.1.1.7-infected subjects are consistent with increased replication capacity and/or less effective clearance by innate or adaptive immune response of B.1.1.7 strain than wild-type. As viral genotype was associated with diverse virological and immunological phenotypes, metrics of viral load, antibody status and infecting strain should be used to define subgroups for analysis of treatment efficacy.