Normal moveout (NMO) velocity is used in seismic data processing to correct the data from the moveout effect. This velocity depends on the medium above the reflector and it is estimated from the adjustment of a hyperbolic function that approximates the reflection time. This approximation is reasonable for media formed by isotropic layers. For deeper exploration targets, which effectively behave as anisotropic media, the NMO velocity estimate from the hyperbolic approximation becomes imprecise. One possibility is the use of non-hyperbolic approximations for the reflection time and deeming the medium to be anisotropic. However, these approximations make the NMO velocity estimation a more complex problem, since the anisotropic parameters are unknown. In this study the NMO velocities for a vertical transverse isotropy medium are estimated using two non-hyperbolic reflection time approaches. For comparing the two methodologies that estimate NMO velocity, a 2-D dataset from Jequitinhonha Basin is used and it presents anisotropic behavior. The results show that this approach produces more consistent results than the conventional approach, which ignores the anisotropy of the medium.