order Born approximation is applied to linearize the integral equations. From these simplified equations, sensitivity functions are derived which define the ability of the crosswell magnetic field measurements to detect small local changes in conductivity. Plots of the spatial variation in sensitivity indicate that vertical resolution is a function of the spatial sampling density in the vertical direction, while the horizontal resolution depends on the vertical separation between the source and receiver. Additional improvement in the horizontal resolution can be provided by making additional measurements with both the source and receiver in the same hole.Further sensitivity analysis is completed in terms of a dimensionless background induction number that is defined by oopl2 where o is the background conductivity and 1 is the source-receiver separation. Analysis of the crosswell sensitivity in terms of this parameter facilitates experimental system design for a given borehole separation and background conductivity, as well as the determination of the operating frequency which falls within an optimal range of oo@ 2.For oopl2 ~1 0 , the scattered fields are very small compared to the primary field, making measurements of the inhomogenous response very difficult in the presence of noise. In addition, a given source-receiver pair is sensitive to a large volume of strata outside of the interwell zone rather than just the region immediately between the probes. The latter property requires that a larger region be considered in the interpretation phase and that the proper 2-D or 3-D model geometry is employed. As the induction number is increased the nature of the sensitivity functions change. For oopl2 > 50 a given source-receiver pair is sensitive only to a "ray-path" shaped zone immediately between the two probes and the scattered fields are on the same order of magnitude as the primary fields. Unfortunately the attenuation at these large induction numbers prevents accurate measurement of the fields for large vertical probe separation due to the dynamic range limitations of the measurement system.The validity of using a 2-D cylindrical model to simulate a generally inhomogenous earth is investigated by comparing the 2-D Born series solution to other 1-D, 2-D and 3-D modeling algorithms. For ooyl2 < 50 the size of the region outside of the wells that is included in the model has a significant effect on the properties of the calculated EM fields with the different model geometries producing significantly different results. At larger induction numbers (owpl2 >50) a given source-receiver pair senses only the medium directly between the probes and the resulting fields are fairly independent of whether a 1-D, 2-D or 3-D model is employed. Unfortunately at these large induction numbers the Born series does not readily converge.To image the subsurface conductivity smcture between two wells, the 2-D cylindrical Born series approximation is incorporated into an iterative Born inversion scheme. The inversion problem is stabiliz...