A prospective evaluation of general medical patients at the University of Alabama at Birmingham was performed in 1976 and repeated in 1988 to determine change in malnutrition prevalence. Plasma folate, plasma ascorbate, weight for height, triceps skinfold, arm muscle circumference, lymphocyte count, albumin, and hematocrit measurements were combined to form a likelihood of malnutrition (LOM) score. The nutritional status of 228 consecutive patients was assessed by the LOM score at admission and at the 14th day of hospitalization and compared with 1976 findings. The same testing methods were used and the same patient diagnoses and demographic characteristics were found in 1988 and 1976. Of the patients staying more than 14 days, the length of stay was the same in 1988 and 1976 (30 days and 31 days, respectively). However, a smaller percentage of patients stayed 2 weeks or longer in 1988 (21% vs 33% in 1976). In 1988, high LOM scores at admission predicted longer lengths of stay and showed a trend toward increased mortality. The 1976 findings also showed that high LOM scores were associated with longer lengths of stay and increased mortality. LOM scores paired from admission to follow-up improved with stay in 1988 and worsened in 1976. The number of patients with high LOM scores at follow-up was lower in 1988 than in 1976 (46% and 62%, respectively). These findings indicate that identification of malnutrition indicators has improved since 1976. However, dietitians should continue to improve the nutrition assessment and intervention process. © 1993 American Dietetic Association.