An inverse proportionality between the specific contact resistance ρc of n-type GaAs and its carrier concentration ND has been shown by a lot of experimental facts. In this paper, a comment is made on the various viewpoints about this phenomenon in the literatures and their shortages are pointed out as well. According to the band structure of ohmic contact, a new model, assuming that ρc consists of two parts (ρ(c1) and ρ(c2)), is put forward. ρ(c1) occurs between the contact metal and its underlying highly doped semiconductor layer (NDc) after alloying. ρc2 is brought about by a barrier appeared due to the concentration difference between the NDc and the active layer ND. If the alloying process is optimized and thus the NDc is very high, then ρ(c1) is very small and ρ(c2) gives the main contribution to ρc. In this case an inverse proportionality between ρc and NDcan be found, if NDc (Nc is the effective state density). When ND> Nc, ρ(c2) can be neglected due to the disappearance of the barrier. In this case, ρc is determined by ρc1, which should depend only on NDc. Based on the above description, a theoretical deduction was carried out and the result not only can explain the experimental data of n-type GaAs ohmic contact very well, but also the experimental facts of p-type Si ohmic contact presented in the literatures. We believe that this model can also be extended to the case of other III-V compound semiconductors, such as p-type GaAs and P-type InP etc.