Previous studies have acknowledged that higher body weight and waist circumference were associated with an increased risk of heart failure. Notably, both body weight and waist circumference can change over time. However, no previous study has investigated the association between combined changes in weight and waist circumference in middle-aged and older adults and incident heart failure.This prospective study included 45 620 middle-aged and older Chinese adults (aged 45-104 years). These participants were free of critical diseases at baseline, including coronary heart disease, stroke, heart failure, atrial fibrillation and cancer. Weight change from 2006-2007 to 2012-2013 was categorized into five groups: excessive weight loss (change < -10%, N = 3943), lesser weight loss (-10% ≤ change < -5%, N = 5890), stable weight (±5%, N = 23 208), lesser weight gain (5% < change ≤ 10%, N = 7153) and excessive weight gain (> 10%, N = 5426). Waist circumference change was categorized into five groups: excessive waist circumference loss (change < -10%, N = 8236), lesser waist circumference loss (-10% ≤ change < -5%, N = 6215), stable waist circumference (±5%, N = 16 953), lesser waist circumference gain (5% < change ≤ 10%, N = 6642) and excessive waist circumference gain (> 10%, N = 7574). Combined changes in weight and waist circumference were divided into 25 groups, i.e., cross-classified combinations derived from the five categories of weight change and five categories of waist circumference change. Incident heart failure cases that occurred from 2012-2013 to December 31, 2022 were recorded. Cox proportional hazards regression models were used to estimate the associations of weight change, waist circumference change or both with heart failure. Multivariate models were stratified by age at risk (in 5-year intervals) and sex, and were adjusted for variables including height, smoking, drinking, educational attainment, occupation, dietary pattern, physical activity, hypertension, fasting blood glucose and total serum cholesterol. In the analysis of weight change, we additionally adjusted for weight at baseline and waist circumference change. Conversely, for the analysis of waist circumference change, adjustments were made for baseline waist circumference and weight change. When examining combined weight and waist circumference changes, adjustments were made for both baseline weight and waist circumference. Additionally, we employed restricted cubic spline analyses to examine the nonlinear associations between changes in weight or waist circumference and heart failure.We identified 1036 heart failure cases during follow-up. The median (interquartile range, IQR) of follow-up time was 9.66 (9.40, 9.96) years. The incidence rate of heart failure was 2.47 cases per 1000 person-years. The median (IQR) age of our participants was 52.1 (46.8, 57.7) years. The proportion of men was 77.9%. The mean (standard deviation) of weight and waist circumference at baseline (baseline, 2006-2007) was 70.1 (10.4) kg and 87.2 (9.0) cm, respectively. Compared with those who kept stable weight, participants in the excessive weight gain group had a higher risk (HR [hazard ratio], 1.27; 95% CI [confidence interval]: 1.03-1.57). Compared with those who kept stable waist circumference, participants in the excessive waist circumference gain group had a higher risk (HR, 1.28; 95% CI: 1.05-1.56), while those in the excessive waist circumference loss group had a lower risk of heart failure (HR, 0.76; 95% CI: 0.64-0.92). Compared with participants with stable weight and waist circumference, those who lost excessive weight and kept stable waist circumference (HR, 1.53; 95% CI: 1.10-2.14), those who lost lesser weight and gained excessive waist circumference (HR, 2.19; 95% CI: 1.38-3.46), and those who gained excessive weight and excessive waist circumference (HR, 1.48; 95% CI: 1.03-2.14) had a higher risk of heart failure. The restricted cubic spline illustrated a U-shaped relation between weight change and incident heart failure (P overall = 0.027, P for non-linear relation = 0.007), whereas a positive linear relation was observed for waist circumference change with incident heart failure (p overall < 0.001, p for non-linear relation = 0.675).Excessive weight gain and waist circumference gain were associated with 27% and 28% higher risk of heart failure, while excessive waist circumference loss was associated with a 24% lower risk of heart failure.© 2025 The Author(s). Journal of Cachexia, Sarcopenia and Muscle published by Wiley Periodicals LLC.