The retention of older respondents in a longitudinal study is of important concern to data quality and representativeness of the target population. Participant non-response has been considered a more serious problem among older persons than among younger ones, since dropouts in longitudinal studies of older adults have been reported to be less healthy, more intellectually impaired, and of lower economic status than continuing respondents. There is concern that the sample of the continuing participants might be biased toward the more functional elderly and may lead researchers to underestimate the mental and physical needs of the general population of the aged. In this study, use of a carefully designed refusal conversion protocol at first follow-up in a longitudinal study of a representative sample of over 4,000 older persons was investigated. Respondents who refused initially did not differ from respondents by age or gender, but were more functionally independent, citing disinterest more frequently as the reason for refusal. The protocol was successful in convincing 43% of these initial refusals to continue participation, but less successful with proxy respondents (27% converted). The findings suggest that most refusals can be converted by telephone at minimal cost, and that it is possible to retain the very old and disabled in longitudinal studies given appropriate field methods. (Author/NB)