Electromagnetic (EM) induction measurements are affected by resistivity and magnetic susceptibility. Thus, inverting EM data for resistivity alone can give misleading models if susceptible effects are strong. An inversion algorithm is presented to simultaneously recover multidimensional distributions of resistivity and susceptibility from various types of loop-loop frequency-domain EM data. The algorithm adopts a staggered-grid finite-difference method for the 3D forward solutions and computes the sensitivities with respect to resistivity and susceptibility from the forward solutions using the reciprocity principle. The algorithm is tested on synthetic data sets from ground-based small-loop, airborne, and Slingram EM surveys. It is shown that the simultaneous inversion of the small-loop EM data collected at a singleheight is unstable and likely to produce unreliable susceptibility models because the effect of susceptibility is nearly independent of the frequency. However, if the data are obtained for multiple heights or different loop configurations, simultaneous inversion can produce more reliable susceptibility and resistivity models even if the data are contaminated by offset errors. It is also shown that although the simultaneous inversion of airborne EM data is relatively stable, adding data obtained at different heights helps to increase the reliability of the resistivity and susceptibility models. Among the loop-loop EM methods discussed here, the Slingram method is relatively insensitive to susceptibility anomalies and thus cannot be used to recover the susceptibility distribution via inversion even if the data are obtained using different loop configurations.