Measures of disability in activities of daily living (ADL) have become important indicators of the health of older persons. One fundamental decision in disability research is constructing or choosing a rating scale to measure ADL disability. Although there is growing consensus in the field on what ADLs to measure, there is little agreement on how to measure ADL disability. This study compares the effect of scales that rate the presence of difficulty, use or human assistance and use of any type of assistance to perform seven different ADLs on prevalence estimates of disability in a probability sample of 1818 adults 70 years of age and older living in the six New England states. Results reveal that different disability rating scales can have a dramatic impact on prevalence estimates of disability in older populations. Measures that used the 'difficulty' scale produced disability estimates from 1.2 to 5 times greater than estimates from the 'human assistance' scale. The effect of rating scales was associated with respondents' age, social factors, and health status. Effects also varied substantially across different ADLs. Researchers need to make careful choices of the disability ratings scales and use caution in drawing inter-study comparisons where different scaling methods were employed.