Three studies to develop and test an instrument to measure hospitalized patients' satisfaction with nursing care are reported. Content validation procedures involved both clinicians and patients (N = 75). An inverse relationship of satisfaction scores to negative mood states demonstrated evidence of construct validity. Factor analytic procedures (N = 664) failed to confirm the existence of the subscales initially conceptualized for this instrument and others in common use. Three new factors were identified: dissatisfaction, interpersonal support, and good impression. The dissatisfaction subscale explained 73.6% of the variance and had a high internal consistency (a = .91); the reliability coefficients of the other subscales also were high (a = .92 and .89). Reliability coefficients for the total instrument in successive testings were .92 (N = 100) and .95 (N = 533).