One of the most influential factors to the complex modulus of a viscoelastic material is the frequency of excitation. In case of estimating such frequency dependent characteristics with longitudinal vibration of a rod-type specimen, it should be considered seriously that how to model the behavior of the specimen and how to derive the complex modulus from the measured frequency response function. In this study, it is investigated how to obtain reasonable material properties from experimental results with respect to these two problems. The strict modeling of the specimen is very difficult, because the boundary conditions of the specimen and lateral motion related to the Poisson's ratio are hard to deal with. Some practical models describing the linear longitudinal vibration of a viscoelastic rod under appropriate assumptions are investigated with respect to available frequency limit on the basis of the exact theory and finite element analysis. Also it is shown that the static model for the bonded effect at specimen ends can be applicable up to the dynamic range used practically. Various techniques have been developed to estimate dynamic mechanical properties of viscoelastic materials, each of which has, of course, its own advantages as well as disadvantages. Non-resonance method is usually used to obtain the complex modulus over wide range of frequency including resonance points. However, the complex modulus obtained by this method often makes the estimations in the anti-resonance frequency regions unreliable because of the measurement noise. In this study, the effects of the measurement errors on estimating the complex Young's modulus by two non-resonance methods; transmissibility and impedance in case of longitudinal vibration of rod-type specimen are studied in the aspect of sensitivity with respect to the measurement noise, and then how to obtain the reliable frequency region for given error bounds is shown. These have been supported by experimental results in case of transmissibility approach.