Introduction Dynamic causal modeling (DCM) based on functional magnetic resonance imaging (FMRI) and magneto-/electroencephalographic (MEEG) data has increasingly extended the understanding of intrinsic brain network dynamics in a variety of functional systems (Friston et al., Curr Opin Neurobiol, 2013). In the motor system, several studies have used DCM to detect causal information flow from secondary to primary motor areas during simple movements and bimanual coordination (Grefkes et al., Neuroimage, 2008). There is also evidence for non-linear cross-frequency interactions among motor areas (Chen et al., J of Neuroscience, 2010; Herz et al., Neuroimage, 2012). As such, data based on MEEG and FMRI provide complementary and synergistic insights into the dynamics of motor networks due to the higher spatial resolution of brain activity by FMRI versus the frequency-resolved coupling revealed by MEEG. A study combining FMRI- and EEG-based DCM for induced responses (DCM-IR) using the same task in the same participants to directly compare network architectures deriving from BOLD response and spectral neuroelectric dynamics has not been conducted so far. Specifically, such an approach is of interest as both modalities share principal similarities in the formulation of the DCM. We hypothesized that DCM based on induced responses and the BOLD-signal would reveal a similar network architecture. Methods To test this, we measured 14 young healthy individuals during a simple isometric hand grip task using FMRI and EEG separately, set up a common model space of six equally plausible models and modeled the coupling parameters within a core motor network during the right hand grip. Results Bayesian Model Selection revealed strongest evidence for a fully connected model in DCM for FMRI and a sparsely connected model in DCM-IR, comparing the individual interregional coupling parameters revealed interesting similarities: First, both modalities showed a significant grip-related increase in facilitatory coupling from the left SMA onto the left M1. Second, the left PMv also exerted a positive coupling onto the left M1 with a significant result for DCM-IR and a trend of significance for DCM for fMRI. Frequency-resolved coupling showed that the information flow from SMA to M1 was a linear alpha to alpha interaction but also a nonlinear cross-frequency interaction between faster oscillations in SMA (18–25 Hz) and full range alpha to beta (9–22 Hz) in left M1. Coupling between PMv and M1 was found from upper alpha (10–13 Hz) to lower beta (14–22 Hz). Conclusion The strategy of informing EEG source space configurations with FMRI location priors, cross-validating basic connectivity maps and then looking at the details of frequency coding allows for a deeper insight into the motor network architecture in the human brain. These results thus undermine the importance of these functional connections and validate the methodological approach of DCM to explore network architecture. Furthermore, extending findings from previous studies showing a covariation of the BOLD-signal and alpha/beta band power (Ritter et al., Human Brain Mapping, 2009; Yuan et al., Neuroimage, 2010), we here present evidence that these two signals not only share similarities in local activation patterns but also in the connectivity domain.