Background. Proteinuria has been related to all-cause mortality, showing regression or progression. However, few studies have focused on the relationship between proteinuria changes and all-cause mortality. The main purpose of this paper is to examine the associations between proteinuria changes and all-cause mortality in people with diabetes or prediabetes. Methods. Dipstick proteinuria at baseline and a 2-year follow-up were determined in the participants attending the Kailuan prospective cohort study. Participants were then divided into three categories: elevated proteinuria, stable proteinuria, and reduced proteinuria. Four Cox proportional hazard models were built to access the relations of proteinuria changes to all-cause mortality, adjusting for other confounding covariates. Results. A total of 17,878 participants were finally included in this study. There were 1193 deaths after a median follow-up of 6.69 years. After adjusting for major covariates and proteinuria at baseline, mortality risk was significantly associated with elevated proteinuria (hazard ratio (HR): 1.54, 95% confidence interval (CI): 1.33–1.79) and reduced proteinuria (HR: 0.70, 95% CI: 0.55–0.89), compared to those with stable proteinuria. Conclusion. Proteinuria changes were independently associated with mortality risk in either diabetic or prediabetic population.