Objectives:We evaluated the accuracy of a newly developed self-completed Drinks Diary in care home residents and compared it with direct observation and fluid intake charts. Design: Observational study. Setting: Residential care homes in Norfolk, UK. Participants: 22 elderly people (18 women, mean age 86.6 years SD 8.6, 12 with MMSE scores <27). Measurements: Participants recorded their own drinks intake over 24 hours using the Drinks Diary while care staff used the homes' usual fluid intake chart to record drinks intake. These records were compared with drinks intake assessed by researcher direct observation (reference method), during waking hours (6am to 10pm), while drinks taken from 10pm to 6am were self-reported and checked with staff. Results: Drinks intake assessed by the Drinks Diary was highly correlated with researcher direct observation (Pearson correlation coefficient r=0.93, p<0.001, mean difference -163ml/day) while few staff-completed fluid charts were returned and correlation was low (r=0.122, p=0.818, mean difference 702ml/day).