In a random sample of 20 hospitals, the availability and consistency of five patient outcome indicators were examined, including medication administration errors, patient falls, occurrence of new decubitus ulcers, nosocomial infections, and unplanned readmission to the hospital. The results indicate that information about only two outcome indicators--medication errors and patient falls--were collected consistently by the sampled hospitals. The findings are discussed in the context of implications for the study of patient outcomes research.