We assembled a data set of permeability measurements from 317 subduction zone and reference site samples worldwide made over nearly 25 years of scientific drilling. This data set allowed us to examine the influence of grain size, structural domain, and measurement type on permeabilities ranging from 10 À21 to 10 À14 m 2 . We found that porosity-permeability behavior is a function of clay-size fraction, which is consistent with previous work. Sediments within the slope, accretionary prism, and fault-zone structural domains are strongly affected by shearing, which alters the permeability behavior with burial. Consolidation, flow-through, and transient pulse decay measurements all provide comparable results. Measurements of horizontal and vertical permeability show significant cm-scale permeability anisotropy (ratio of horizontal to vertical permeability >10) in the slope and accretionary prism structural domains, further indicating shear deformation in these domains. Laboratory consolidation trends match large-scale (10 2 m) field trends in structural domains with negligible shear, but tend to underestimate the rate of permeability reduction with porosity loss where shear is significant. Comparison with downhole measurements shows that permeability is controlled by higher-permeability (>10 À15 m 2 ) layers at the meter to tens of meters scale, while wireline formation tester measurements closely match laboratory results. Sediments from the underthrust and reference structural domains exhibit similar porosity-permeability trends, which suggests that shallow subduction (total burial <1 km) does not significantly alter the porosity-permeability behavior of incoming sediments. Comparison with measurements of deeper analog data from 14 passive-margin samples show that porosity-permeability trends are maintained through burial and diagenesis to porosities <10%, suggesting that behavior observed in shallow samples is informative for predicting behavior at depth following subduction.