Abstract Objectives To examine the association of protein intake with frailty progression in very old adults. Design The Newcastle 85+ study, a prospective longitudinal study of people aged 85 years old in Northeast England and followed over 5 years. Setting and Participants 668 community-dwelling older adults (59% women) at baseline, with complete dietary assessment and Fried frailty status (FFS). Measures Dietary intake was estimated with 2 × 24-h multiple pass recalls at baseline. FFS was based on five criteria: shrinking, physical endurance/energy, low physical activity, weakness and slow walking speed and was available at baseline and 1.5, 3 and 5 years. The contribution of protein intake (g/kg adjusted body weight/day g/kg aBW/d) to transitions to and from FFS (robust, pre-frail and frail) and to death over 5 years was examined by multi-state models. Results Increase in one unit of protein intake (g/kg aBW/d) decreased the likelihood of transitioning from pre-frail to frail after adjusting for age, sex, education and multimorbidity (hazard ratios HR: 0.44, 95% confidence interval CI: 0.25–0.77) but not for the other transitions. Reductions in incident frailty were equally present in individuals with protein intake ≥0.8 (HR: 0.60, 95% CI: 0.43–0.84) and ≥1 g/kg aBW/d (HR: 0.63, 95% CI: 0.44–0.90) from 85 to 90 years. This relationship was attenuated after adjustment for energy intake, but the direction of the association remained the same (e.g. g/kg aBW/d model: HR: 0.71, 95% CI: 0.36–1.41). Conclusion High protein intake, partly mediated by energy intake, may delay incident frailty in very old adults. Frailty prevention strategies in this age group should consider adequate provision of protein and energy.