We aimed to determine whether serum leptin levels are predictive of the occurrence of healthcare-associated infections (HAIs) in hospitalized older patients. In a prospective cohort, 232 patients had available data for leptin and were monitored for HAIs for 3 months. Admission data included comorbidities, invasive procedures, the Mini Nutritional Assessment (MNA), BMI, leptin, albumin and C-reactive protein levels, and CD4 and CD8 T-cell counts. Multivariate logistic regression modelling was used to identify predictors of HAIs. Of the 232 patients (median age: 84.8; females: 72.4%), 89 (38.4%) experienced HAIs. The leptin level was associated with the BMI (p < 0.0001) and MNA (p < 0.0001) categories. Women who experienced HAIs had significantly lower leptin levels than those who did not (5.9 μg/L (2.6–17.7) and 11.8 (4.6–26.3), respectively; p = 0.01; odds ratio (OR) (95% confidence interval): 0.67 (0.49–0.90)); no such association was observed for men. In a multivariate analysis of the women, a lower leptin level was significantly associated with HAIs (OR = 0.70 (0.49–0.97)), independently of comorbidities, invasive medical procedures, and immune status. However, leptin was not significantly associated with HAIs after adjustments for malnutrition (p = 0.26) or albuminemia (p = 0.15)—suggesting that in older women, the association between serum leptin levels and subsequent HAIs is mediated by nutritional status.