In a study to identify factors that cause decubitus ulcers and to describe characteristics of patients who develop them, 26 nonambulatory patients in two health care facilities were evaluated two times a week for four weeks or until a decubitus ulcer formed. A point-biserial correlation for age and skin breakdown, a phi correlation for 20 other variables, and a stepwise regression, using information from the correlations, produced a profile of a patient likely to develop these ulcers: a thin, febrile male with an infection, who may be receiving corticosteroid therapy. Use of the 21 variables, the study findings revealed, could predict decubitus ulcer formation with 98.26 per cent accuracy.