Abstract: In this paper the exact Bayesian and Fiducial limits for coefficient of variation of normal distribution are derived. When we take the noninformative prior distribution on the normal parameters, the Bayesian limits for Coefficient of Variation are equal to the Fiducial limits. They can be expressed by the nonlinear equation for CDFs of the noncentral t-distribution and Beta distribution. An approximates Bayesian limits having enough accuricy are presented to meet practical engineering needs. It can be applied directly. When it is used as original ranges to solve the exact limits by iterative method, the com, putation can be decreased.