Our objective was to estimate the cost of inpatient malnutrition conditional upon admitting diagnosis and recorded nutritional treatment. We analysed an anonymised administrative data set for inpatients treated in hospitals in Victoria, Australia 2003-2004. The data set included patient-level cost, clinical markers, demographic and episode variables. The data set was analysed to identify codes, which indicated the presence of malnutrition or its treatment. Linear regression was used to ascertain to what extent malnutrition affected the cost of admission. Controlling for the underlying condition and any treatment administered, recorded malnutrition is estimated to add AU $1,745 per admission. The total cost of coded malnutrition to the Victorian public hospital system in 2003-2004 was estimated to be least AU $10.7 million. Only 1.87% of inpatients were coded as malnourished. As administrative data are known to underreport the prevalence of malnutrition, our estimate represents a credible lower boundary on the true cost of inpatient malnutrition.