Background There are no data on SARS-CoV-2 seroprevalence in Africa though the COVID-19 epidemic curve and reported mortality differ from patterns seen elsewhere. We estimated the anti-SARS-CoV-2 antibody prevalence among blood donors in Kenya. Methods We measured anti-SARS-CoV-2 spike IgG prevalence by ELISA on residual blood donor samples obtained between April 30 and June 16, 2020. Assay sensitivity and specificity were 83% (95% CI 59, 96%) and 99.0% (95% CI 98.1, 99.5%), respectively. National seroprevalence was estimated using Bayesian multilevel regression and post-stratification to account for non-random sampling with respect to age, sex and region, adjusted for assay performance. Results Complete data were available for 3098 of 3174 donors, aged 15-64 years. By comparison with the Kenyan population, the sample over-represented males (82% versus 49%), adults aged 25-34 years (40% versus 27%) and residents of coastal Counties (49% versus 9%). Crude overall seroprevalence was 5.6% (174/3098). Population-weighted, test-adjusted national seroprevalence was 5.2% (95% CI 3.7, 7.1%). Seroprevalence was highest in the 3 largest urban Counties; Mombasa (9.3% [95% CI 6.4, 13.2%)], Nairobi (8.5% [95% CI 4.9, 13.5%]) and Kisumu (6.5% [95% CI 3.3, 11.2%]). Conclusions We estimate that 1 in 20 adults in Kenya had SARS-CoV-2 antibodies during the study period. By the median date of our survey, only 2093 COVID-19 cases and 71 deaths had been reported through the national screening system. This contrasts, by several orders of magnitude, with the numbers of cases and deaths reported in parts of Europe and America when seroprevalence was similar.