PURPOSE: To assess relative reliability and measurement error of the Graded Cycling Test (GCT) with the Talk Test (TT) for patients with cardiac disease. METHODS: Patients (N = 64; women, n = 30) with ischemic heart disease performed the GCT with the TT twice in 1 day. Every minute the patient recited a text passage of 30 words. The question "Are you still able to speak comfortably?" was then answered: "Yes" (TT+), "Not sure" (TT+/-), or "No" (TT-). Two physiotherapists (PTs) rated when the speaking ability was affected by ventilation. Relative reliability: intraclass correlation coefficient (ICC)2.1 and absolute reliability: standard error of measurement with 95% CI (SEM95) and smallest real difference (SRD) were calculated. RESULTS: Intraclass correlation coefficient values of 0.90, 0.91, and 0.90 were observed for TT+, TT+/-, and TT-, respectively. Physiotherapist ICCs ranged between 0.81 and 0.88. SEM95 ranged between 17.2 and 18.3 watts (W), with corresponding SRD values between 24.4 and 25.9 W for the patient ratings. The PT ratings ranged between 15.8 and 21.4 W (SEM95) and between 22.3 and 30.3 W (SRD). CONCLUSIONS: The TT, combined with the GCT, was well tolerated by patients with cardiac disease in clinical settings with small measurement error and an excellent relative reliability. These results apply to both patient-rated and PT-rated TT.