In a unified viewpoint in quantum channel estimation, we compare the Cramér-Rao and the mini-max approaches, which gives the Bayesian bound in the group covariant model. For this purpose, we introduce the local asymptotic mini-max bound, whose maximum is shown to be equal to the asymptotic limit of the mini-max bound. It is shown that the local asymptotic mini-max bound is strictly larger than the Cramér-Rao bound in the phase estimation case while the both bounds coincide when the minimum mean square error decreases with the order O( 1 n ). We also derive a sufficient condition for that the minimum mean square error decreases with the order O( 1 n ).Simpler estimation scheme of quantum channel different analyses were reported concerning asymptotic behavior of MSE. As the first case, in the estimations of depolarizing channels and Pauli channels, the optimal MSE behaves as O( 1 n ) [1,3]. As the second case, in the estimation of unitary, the optimal MSE behaves as O( 1 n 2 ) [4,5,6,7,8,9,10,11]. In the second case, two different types of results were reported: One is based on the Cramér-Rao approach [4,5]. The other is based on the mini-max approach [6,7,8,9,10,11].The Cramér-Rao approach is based on the notion of locally unbiased estimator, and allows one to give a simple lower bound (the Cramér-Rao bound) to the mean square error (MSE) at a given point. The mini-max approach aims to minimize the maximum of the MSE over all possible values of the parameter. The mini-max approach is more meaningful than the Cramér-Rao approach because the true value of the parameter is unknown. So, the Cramér-Rao bound is just Comparison between Cramer-Rao and mini-max approaches 3 a lower bound in general, while it can be asymptotically acheived in the case of quantum state estimation with the n copies of the unknown state. However, the Cramér-Rao bound has been considered so many times in the literatures [1,3,4,5]. The reason seems that computing the mini-max bound is a much harder problem. Indeed, there are only very few examples of calculations of the mini-max bound, and most of them are in the compact group covariant setting, in which, as was shown by Holevo [15], the mini-max bound coincides with the Bayesian average of the MSE over the normalized invariant measure. So, many researchers [6,7,8,9,10,11] applied this approach to quantum channel estimation under the group covariance. As the most simple case for unitary estimation, phase estimation has been treated with the mini-max approach by several papers [9,10,11], and it has been shown that the minimum MSE that behaves as O( 1 n 2 ). On the other hand, the Cramér-Rao approach suggests the noon state as the optimal input [36,37] in phase estimation. The later estimation scheme was experimentally demonstrated in the case of n = 4 [32,33]and n = 10 [34]. Also, another group [31] experimentally demonstrated an estimation protocol concerning the group covariant approach proposed by [35]. In phase estimation with group covariant framework, the asymptotic minimum MSE behaves as ...