Abstract Purpose Our aim was to quantify the mortality from COVID-19 and identify any interactions with frailty and other demographic factors. Methods Hospitalised patients aged ≥ 70 were included, comparing COVID-19 cases with non-COVID-19 controls admitted over the same period. Frailty was prospectively measured and mortality ascertained through linkage with national and local statutory reports. Results In 217 COVID-19 cases and 160 controls, older age and South Asian ethnicity, though not socioeconomic position, were associated with higher mortality. For frailty, differences in effect size were evident between cases (HR 1.02, 95% CI 0.93–1.12) and controls (HR 1.99, 95% CI 1.46–2.72), with an interaction term (HR 0.51, 95% CI 0.37–0.71) in multivariable models. Conclusions Our findings suggest that (1) frailty is not a good discriminator of prognosis in COVID-19 and (2) pathways to mortality may differ in fitter compared with frailer older patients.