Fluid identification and fracture discrimination play an important role in the exploration and development of an oil-bearing fractured reservoir. The most common fluid indicator in fractured reservoirs, the normal-to-shear fracture compliance ratio, is influenced by the fluid content and the fracture intensity. To reduce the ambiguities in the discrimination of the fluid and fracture parameters, we have aimed to extend the scattering theory to implement the fluid identification and fracture detection by incorporating the azimuthal data in an oil-bearing fractured reservoir via the proposed Bayesian amplitude variation with offset and azimuth (AVOAz) inversion approach. The background medium is, as far as the scattering theory is concerned, an isotropic medium without fractures, and the fractured medium is corresponding to a perturbed medium. The elastic parameters of a saturated anisotropic medium can be parameterized as a perturbation over a homogeneous isotropic background medium. We used the scattering theory to derive a generalized AVOAz approximation that provided the iterative estimates of hydrocarbon fluid indicator, shear modulus, density, and fracture weaknesses in a Bayesian scheme. The inversion algorithm is based on a convolutional model and a weak-contrast and small-weakness PP-wave reflection coefficient. The approach is applied to an oil-bearing field data set from a fractured marlstone reservoir. We observe that reasonable estimates of fluid indicator and fracture weaknesses are inverted, which can be used to perform the discrimination of fluid and fracture parameters. We conclude that the proposed approach provides us a potentially powerful tool to estimate the reservoir fluid and fracture properties in a more straightforward and efficient manner than those previous methods.