We recorded the survival of 141 patients assessed for radical cystectomy, which included cardiopulmonary exercise testing. The median Kaplan-Meier survival estimates were: 1540 days for the whole cohort; 2200 days after cystectomy scheduled (n = 108); and 843 days without surgery. The mortality hazard remained double that expected for a matched general population, but survival was better in patients scheduled for surgery than those who were not: the mortality hazard ratio (95%CI) after cystectomy was 0.43 (0.26-0.73) the mortality hazard without surgery, p = 0.001. The mortality hazard ratios for the three-variable Bayesian Model Averaging survival model for all 141 patients were: referral for surgery (0.5); haemoglobin concentration (0.98); and efficiency of carbon dioxide output (1.05). Efficiency of carbon dioxide output was the single variable in the postoperative model (n = 108), mortality hazard 1.08 (per unit increase). The ratio of observed to expected peak oxygen consumption associated best with mortality in 33 patients not referred for surgery, hazard ratio 0.001. Our results can inform consultations with patients with invasive bladder cancer and suggest that interventions to increase fitness and haemoglobin may improve survival in patients who do and who do not undergo radical cystectomy.