The advent of clinical trials of disease-modifying agents for neurodegenerative disease highlights the need for evidence-based endpoint selection. Here we report the longitudinal PROSPECT-M-UK study of progressive supranuclear palsy, corticobasal syndrome, multiple system atrophy and related disorders, to compare candidate clinical trial endpoints. In this multicentre United Kingdom study, participants were assessed with serial questionnaires, motor examination, neuropsychiatric and magnetic resonance imaging assessments at baseline, six and twelve-months. Participants were classified by diagnosis at baseline and study end, into Richardson syndrome, progressive supranuclear palsy-subcortical (progressive supranuclear palsy-parkinsonism and progressive gait freezing subtypes), progressive supranuclear palsy-cortical (progressive supranuclear palsy-frontal, progressive supranuclear palsy-speech-and-language, and progressive supranuclear palsy-corticobasal syndrome subtypes), multiple system atrophy-parkinsonism, multiple system atrophy-cerebellar, corticobasal syndrome with and without evidence of Alzheimer’s disease pathology and indeterminate syndromes. We calculated annual rate of change, with linear mixed modelling, and sample sizes for clinical trials of disease modifying agents, according to group and assessment type. Two hundred forty-three people were recruited (117 progressive supranuclear palsy, 68 corticobasal syndrome, 42 multiple system atrophy and 16 indeterminate; 138 [56.8%] male; age at recruitment 68.7 ± 8.61 years). One hundred fifty-nine completed six-month assessment (82 progressive supranuclear palsy, 27 corticobasal syndrome, 40 multiple system atrophy and 10 indeterminate) and 153 completed twelve-month assessment (80 progressive supranuclear palsy, 29 corticobasal syndrome, 35 multiple system atrophy and 9 indeterminate). Questionnaire, motor examination, neuropsychiatric and neuroimaging measures declined in all groups, with differences in longitudinal change between groups. Neuroimaging metrics would enable lower sample sizes to achieve equivalent power for clinical trials than cognitive and functional measures, often achieving N < 100 required for one-year two-arm trials (with 80% power to detect 50% slowing). However, optimal outcome measures were disease specific. In conclusion, phenotypic variance within progressive supranuclear palsy, corticobasal syndrome and multiple system atrophy is a major challenge to clinical trial design. Our findings provide an evidence base for selection of clinical trial endpoints, from potential functional, cognitive, clinical or neuroimaging measures of disease progression.