An analysis of seismic anisotropy at a BHP mining site in the Southern Sydney Basin by combined use of crosshole and vertical seismic profiling (VSP) data is presented. The upper 250 m in this area is highly heterogeneous and has a major impact on the analysis of P-wave traveltimes. It is shown that using P-wave information solely would not, at least in this case, lead to any reasonable estimate of the elastic constants, in particular C 13 , even if the measurements contained a full range of incident angles. However, if the measurements of SV-waves are available, even over a small range of incident angles, then C 13 is determined more accurately. P-wave velocities measured in the vertical and horizontal directions show that anisotropy is present in the area. Additional measurements, along different incident angles, indicate that the rock down to 500 m depth is predominantly transversely isotropic (TI) with a vertical axis of symmetry. The P-wave anisotropy can be approximated as elliptical. Using the elastic constants estimated from the data analyses, synthetic seismograms for heterogeneous TI media were generated. Comparison of the seismic modeling with real crosshole data shows that it is necessary to include both fault zones and gas accumulations in the model to qualitatively match the real data. By using SV-waves in the multioffset VSP data, reflectors are mapped more accurately than by using P-waves, even under the assumption of isotropy and in the presence of heterogeneity. Mapping of converted P-SV waves by a straight ray approach also produced better results than the corresponding isotropic P-wave mapping. Inclusion of elliptical anisotropy into Kirchhoff migration resulted in better P-wave images than using an isotopic migration code. We conclude that both P-wave VSP multioffset mapping and tomographic inversion methods need to account for anisotropy to be accurate in this area, while SV-waves may be handled using isotropic codes. The same is true for crosshole and surface seismic data.