A method for designing a controlled-source electromagnetic survey that will discriminate between a defined deep marginal-interest reservoir (2) and specified false positive resistivity structures of concern (3, 4, 5). A reservoir model and a false positive model are constructed for each false positive scenario. The resistivity of the false positive model may be tuned to give electromagnetic data similar enough to the reservoir model when forward modeled that any differences fall in the model null space. A null-space discriminating ratio (“NSDR”) is defined, for example as the peak normalized difference of the two related modeled electromagnetic field data sets. An area coverage display of NSDR values (6) allows determination of such additional data as may be needed to distinguish the false positive body, and a survey design is developed accordingly (7). Reduction of the number of variables affecting the area coverage displays is a key feature of the method.