For perforated natural completions, well productivity is dependent on the depth of the perforation tunnels extending beyond the drilling damage (all other things being equal). It is therefore important to accurately predict perforation depth at downhole conditions, in order to enable accurate prediction of well performance.Most industry penetration models in use today are based largely on laboratory experiments conducted during the 1960's through the 1990's. Over the past decade or so, it has been observed that the accuracy of these models has not kept up with the true downhole performance of modern shaped charge perforators. Furthermore, different models can give quite different predictions of penetration performance of the same perforating system, in the same downhole environment.To address this situation, the authors have recently conducted an extensive series of laboratory experiments, the results of which are enabling improved prediction of penetration performance in the field. Several hundred modern charges of different sizes have been shot into multiple rock types under simulated downhole stress conditions.The primary conclusions of this work include: (1) historical penetration models tend to overpredict penetration at downhole conditions; (2) some industry models overpredict to a greater extent than others; (3) the discrepancy is partly due to the industry's continued reliance on performance into surface concrete targets. In addition, we observe that historical models tend to treat all charges equally, imposing the assumption that increased performance in one target (i.e. surface concrete) always and everywhere ensures increased performance in all targets (i.e. various stressed rocks). While this intuitive assumption is proven true as the "rule" on average, we find some significant exceptions. Although these exceptions complicate predictive modeling efforts, they do suggest opportunities to optimize charges for certain targets preferentially over others.Our conclusions have led to ongoing work including (1) the development of improved penetration models, which rely exclusively on stressed rock, rather than unstressed concrete, performance; (2) the development of perforators optimized for downhole conditions, rather than for surface concrete.