AVO inversion of prestack seismic data, constrained by geologic knowledge and rock-physics modeling, is an essential technology that can greatly reduce interpretation risk during exploration. However, in spite of its great potentials, this technology suffers from a wide range of pitfalls and many uncertainties. In this study, we demonstrate the use of simultaneous AVO inversion in an area of the Norwegian Sea where several proven discoveries with strong AVO anomalies are present, all of which are structural traps. The inversion successfully delineated these discoveries, and a pure blind test of a prospect on a structural high correctly predicted the presence of hydrocarbons. The AVO inversion also gave support for a stratigraphic trap in a graben setting between the prolific structural highs. However, this prospect turned out to be a failure, as the observed AVO anomaly was false. Postdrill analysis showed that a thin, very hard, calcareous event right above the target had created a refraction that interfered with the target horizon below, creating a false AVO class II to III anomaly. The target interval was predominantly a very soft, thick, and immature organic-rich shale, which normally should show a class IV AVO anomaly. Moreover, the low-frequency model used in the inversion was derived from wells on structural highs combined with interval velocities and turned out to not be representative in the graben areas. The key learnings from this study are not to use angle ranges exceeding about 40° during AVO inversion in this area and, in general, to be careful extrapolating the low-frequency model away from well control, even when interval velocities are used, especially in areas with complex geology. More integration between geology and geophysics is necessary during all steps of the AVO inversion to avoid such pitfalls.