As part of a study of patient-centered care outcomes that requires the ability to be interviewed by telephone after hospitalization, 4,600 adult patients on 118 medical-surgical units in 17 midwestern hospitals selected by stratified random sampling were identified as potential subjects. This article describes response-rate differences and reasons for nonparticipation by gender, age, ethnicity, and race. Issues related to understanding and reporting response rates, reducing losses due to ineligibility, dealing with refusals, and tailoring survey approaches to special populations are discussed.