Abstract A close scrutiny of core data from a producing carbonate field resolved an apparent difference between permeability evident from core plug analysis and from testing of wells in the field. This was attributed to a bias inherent in the data collection, particularly as the samping techniques avoided the most permeable features. Instead of invoking a pervasive fracture network in the static geological model, which might compromise a planned water injection campaign, it is now evident that the permeability mismatch can be satisfactorily modeled with a horizontal multiplier, and that water injection should result in a significant increase in ultimate oil recovery.