Thermal behavior of isoscalar (τ =0) and isovector (τ =1) proton-neutron (pn) pairing energies at finite temperature are investigated by the shell model calculations. These pn pairing energies can be estimated by double differences of "thermal" energies which are extended from the double differences of binding energies as the indicators of pn pairing energies at zero temperature. We found that the delicate balance between isoscalar and isovector pn pairing energies at zero temperature disappears at finite temperature. When temperature rises, while the isovector pn pairing energy decreases, the isoscalar pn pairing energy rather increases. We discuss also the symmetry energy at finite temperature. The proton-neutron (pn) pairing energies have become one of hot topics in the study of the nuclear structure for proton-rich nuclei. In particular, interests are increasing in studying isovector (τ =1) and isoscalar (τ =0) pn pairing energies in medium mass N = Z nuclei produced at the radioactive nuclear beam facilities. The study of pn pairing energies is also important in the astrophysical context. These nuclei lie along the explosive rp-process nucleosynthesis path and the nuclear properties such as masses, halflives, and isomers have a strong influence on modeling the rp-process and identifying possible nucleosynthesis sites. Odd-odd N = Z nuclei are an ideal experimental laboratory for the study of pn pairing energies. It is well known that the lowest τ = 0 and τ = 1 states compete for the ground state changing the sign of the energy difference E τ =1 − E τ =0 in odd-odd N = Z nuclei, while all even-even N = Z nuclei have the τ = 0 ground states. Several authors [1,2,3,4,5,6,7] already pointed out that this degeneracy in odd-odd N = Z nuclei reflects the delicate balance between the symmetry energy and the like-nucleon neutron-neutron (nn) (or proton-proton (pp)) pairing energy. On the other hand, it has recently been shown that this degeneracy is attributed to competition between the isoscalar and isovector pairing energies [9,10, 25].It has been recently reported [11,12] that the canonical heat capacities extracted from observed level densities in 162 Dy, 166 Er and 172 Yb display the S shape with a peak around T ≈ 0.5 MeV, which is interpreted as the breaking of like-nucleon J = 0 pairs because the BCS critical temperature corresponds to T c ≈ 0.57∆ n (T = 0) ≈ 0.5 MeV, where the like-nucleon pairing gap ∆ n (T = 0) is calculated at zero temperature by the BCS theory. Thus it seems that the S shape is a signature of pairing transition at the critical temperature. For the finite Fermi system like a nucleus, however, since the nuclear radius is much smaller than the coherence length of the Cooper pair, statistical fluctuations beyond the mean field in the BCS theory become large. The fluctuations smooth out the sharp phase transition, and then the like-nucleon pairing gap ∆ n does not quickly become zero at the BCS critical temperature but decreases with increasing temperature. There are many approaches to treat...