Objective: to develop a measure of activities of daily living appropriate for use in assessing the presence of dementia in illiterate rural elderly people in India. Design: identification of relevant items, pre-testing of items and refinement of administrative procedures and scoring in four successive groups of 30 subjects each, pilot testing in a group of 100 subjects comparable to those for whom the measure is intended, administration to a representative sample of 387 people aged 55 and older, and assessment of the reliability of the final measure. Setting and subjects: age-stratified random sample of older men and women in rural areas of Ballabgarh, Northern India Results: the original pool of 35 items covering mobility, instrumental and personal care activities was reduced to an 11-item unidimensional scale (to which an additional item on mobility was added) with internal consistency (Cronbach's a) = 0.82, perfect inter-and intra-rater reliability, test-retest reliability (intraclass correlation) = 0.82 (any disability) and 0.92 (unable to perform for 'mental' reasons). Women, older subjects, the totally illiterate and subjects with poorer cognitive function performed significantly more poorly (P Յ 0.02 for all). Product: a brief, reliable and valid activities of daily living measure, with norms, which is appropriate for use in assessing dementia in illiterate rural elderly people in India.