Abstract Background Using Newcastle 85+ Study data, we investigated transitions between frailty states from age 85 to 90 years and whether multi-morbidities and socioeconomic status (SES) modify transitions. Methods The Newcastle 85+ Study is a prospective, longitudinal cohort study of all people born in 1921 in Newcastle and North Tyneside. Data included: a multidimensional health assessment; general practice record review (GPRR) and date of death. Using the Fried phenotype (participants defined as robust, pre-frail or frail), frailty was measured at baseline, 18, 36 and 60 months. Results Frailty scores were available for 82% (696/845) of participants at baseline. The prevalence of frailty was higher in women (29.7%, 123/414) than men (17.7%, 50/282) at baseline and all subsequent time points. Of those robust at baseline, 44.6% (50/112) remained robust at 18 months and 28% (14/50) at age 90. Most (52%) remained in the same state across consecutive interviews; only 6% of the transitions were recovery (from pre-frail to robust or frail to pre-frail), and none were from frail to robust. Four or more diseases inferred a greater likelihood of progression from robust to pre-frail even after adjustment for SES. SES did not influence the likelihood of moving from one frailty state to another. Conclusions Almost half the time between age 85 and 90, on average, was spent in a pre-frail state; multi-morbidity increased the chance of progression from robust and to frail; greater clinical intervention at the onset of a first chronic illness, to prevent transition to multi-morbidity, should be encouraged.