Background: Long-staying medical inpatients carry a significant burden of acute and chronic illness. Prediction of their in-hospital and longer-term mortality risk is important. Aim: The aim of this study was to determine to what extent creatinine variability predicts in-hospital and 1-year mortality in inpatients. Design: Retrospective cohort analysis. Methods: Patients were included if aged 18 years or older and if admitted for 7 days or longer. The main outcome variables were mortality in hospital and after discharge. Results: Increasing age, the presence of heart failure and a reduced estimated glomerular filtration rate (eGFR) on admission (<60 ml/min/1.73 m2) all associated with death risk (both in hospital and within a year of discharge). The creatinine change was related to mortality risk for the patient whilst in hospital and within 1 year after discharge independently of these other factors. The threshold of creatinine change, above which the in-hospital mortality rose significantly was 25 μmol/l (P < 0.001). A creatinine change of >10 μmol/l predicted significantly higher mortality within a year of discharge (P < 0.001). Every 5 μmol/l change in creatinine was associated with an in-hospital mortality increase of 3% (P < 0.001) and a 1-year mortality increase of 1% (P < 0.007).Conclusions: Patients with a creatinine rise or fall of >10 μmol/l during admission are at higher risk of death after discharge than those with more stable creatinine. These patients therefore merit further attention that might include more focused nutritional assessment, cardiovascular risk factor management or advance care planning.