This work investigates the phenomena of self-heating, also called intrinsic heating, and thermoelastic coupling during non-linear dynamic mechanical fatigue testing via surface temperature measurement coupled with the mechanical behavior of polymers. Static tensile tests and dynamic strain controlled fatigue tests under tension/tension were performed at a frequency of ω1/2π = 5 Hz, as well as in the low cycle fatigue regime at ω1/2π = 0.2 Hz, on six polymers: high density polyethylene (HDPE), low density polyethylene (LDPE), ultra high molecular weight polyethylene (UHMWPE), polyamide 6 (PA6), and two grades of polypropylene (PP).In dynamic testing, the surface temperature rises to a plateau value (ΔT) when an equilibrium between the viscous/plastic dissipated energy and heat convection is reached. Power-law correlations were found between the strain amplitude (ε0) and ΔT, as well as between ε0 and the calculated dissipated energy density (Wdiss,p) obtained from the mechanical stress response, with similar exponents for both correlations. Thermoelastic coupling is firstly investigated in uniaxial tension, revealing a linear relation between the strain rate and the rate of temperature decrease, which is more distinct with decreasing polymer chain mobility. In dynamic fatigue testing, the surface temperature was found to oscillate with an amplitude T1, which was analyzed via Fourier transform. A direct relation between T1 and ε0 at small deformations was observed. At large strain amplitudes, T1 (ε0) follows a similar trend as the complex modulus E*(ε0). At low frequencies and large strain amplitudes, additional higher harmonics at two (T2) and three (T3) times the fundamental frequency were also detected as fingerprints of plastic deformation, resulting in additional heat dissipated during the loading half cycle. From the results obtained, the advantages of the calculated dissipated energy density over the surface temperature analysis was analyzed to predict the fatigue behavior. This analysis is believed to be valid for all materials due to the mathematical/physical principles involved. The results are thus expected to hold for other materials such as composites, rubbers, ceramics and metals.