The incidence of heart failure (HF) increases in patients with chronic kidney disease (CKD). Factors that could predict patients with CKD who are at high risk for developing HF should be identified. We analysed clinical parameters and stress/rest myocardial perfusion imaging (MPI) findings derived from 499 patients with CKD by the Japanese Assessment of Cardiac Events and Survival Study by Quantitative Gated SPECT 3 (J-ACCESS 3) to clarify predictors of new-onset HF. Forty-one patients with congestive HF in the J-ACCESS 3 database were followed up for three years. Multivariable Cox hazards models selected haemoglobin (hazard ratio [HR] 0.809; 95% confidence interval [CI] 0.679–0.964), summed stress score (HR 1.082; 95% CI 1.016–1.151) and left ventricular ejection fraction (HR 0.970; 95% CI 0.949–0.992) as independent predictors of new-onset HF. Haemoglobin combined with summed stress scores and ejection fraction had the greatest incremental prognostic value over any one or more combined factors (global χ², 29.9). Anaemia, stress-induced myocardial ischaemia, and left ventricular contraction are independent predictors of risk of new-onset HF in patients with CKD. Stress/rest MPI provides additional information with which to identify patients with CKD at greater risk of new-onset HF.