Background The influence of ethnicity and socioeconomic status on dementia risk remain incom- pletely understood. We addressed this issue using data from over a million people in the diverse population of East London (<50% White and >50% in the most deprived UK quintile). Methods We identified 4137 cases of dementia, and matched on age and gender to controls with ratio 1:4. We calculated odds ratios (ORs) for ethnicity and Index of Multiple Deprivation (IMD), before and after adjustment for established risk factors (diabetes, hypertension, smoking, BMI, depression and hearing loss). We calculated weighted population attributable fractions (PAF) for each factor. Results Risk of dementia was higher for Black and South Asian groups than White (ORs Black 1.43, South Asian 1.17). Risk of dementia was reduced in all IMD quintiles relative to most deprived (ORs 2nd 0.71, 3rd 0.52, 4th 0.71, 5th 0.59). The effects of ethnicity and deprivation persisted after adjusting for known risk factors. Weighted PAFs for modifiable risk factors were notably higher in this popu- lation for depression (9.2%) and diabetes (6.2%) than those estimated in the Lancet Commission meta-analyses (4% and 1.2% respectively). Conclusions Ethnicity and socioeconomic deprivation are important determinants of dementia risk, with effects larger than many of the more established risk factors. Depression and diabetes are of greater relative importance, and should be prioritised as targets for dementia prevention efforts in more diverse and deprived populations. drcharlesmarshall@gmail.com