Little is known about heart rate (HR) kinetics during exercise in the very heavy (VH) and severe (SE) intensity domains. The objective of this study was to describe mathematically the HR kinetics during exercise performed in these intensity domains and to compare the parameters derived from these models between situations of high (HCHO), low (LCHO) and control (C) carbohydrate availability. Twelve men performed three trials to exhaustion in the VH or SE domains after diet manipulation with HCHO, LCHO and C. The VH intensity was ∆LW75% (75% of the difference between VO2max and LL2) and SE was 115% of VO2max identified in a previous incremental test (20 W/3 min). HR responses were mathematically fitted by mono- and biexponential functions. In the VH domain, the residual sum of squares (RSS) obtained with the biexponential model was significantly lower than that obtained with the monoexponential model (P < 0.05). In the SE domain, no significant difference in RSS was observed between the mathematical models (P > 0.05). In the VH domain, there were no significant differences in biexponential parameters between the HCHO, LCHO and C conditions. In the SE domain, there were no significant differences in monoexponential parameters between the HCHO, LCHO and C conditions, although the time constant of the monoexponential model was significantly reduced in LCHO when compared to HCHO (51.5 ± 26.4 vs 65.4 ± 34.1 s; P < 0.05). The bi- and monoexponential mathematical models seem to be the best description of HR responses during exercise performed in the HV and SE intensity domains, respectively. In addition, carbohydrate availability only seems to affect HR kinetics during exercise performed at SE intensity.