For the first time in research in humans, we used simultaneous icEEG-fMRI to examine the link between connectivity in haemodynamic signals during the resting-state (rs) and connectivity derived from electrophysiological activity in terms of the inter-modal connectivity correlation (IMCC). We quantified IMCC in nine patients with drug-resistant epilepsy (i) within brain networks in ‘healthy’ non-involved cortical zones (NIZ) and (ii) within brain networks involved in generating seizures and interictal spikes (IZ1) or solely spikes (IZ2). Functional connectivity ( h 2 ) estimates for 10 min of resting-state data were obtained between each pair of electrodes within each clinical zone for both icEEG and fMRI. A sliding window approach allowed us to quantify the variability over time of h 2 (v h 2 ) as an indicator of connectivity dynamics. We observe significant positive IMCC for h 2 and v h 2 , for multiple bands in the NIZ only, with the strongest effect in the lower icEEG frequencies. Similarly, intra-modal h 2 and v h 2 were found to be differently modified as a function of different epileptic processes: compared to NIZ, h BOLD 2 was higher in IZ1, but lower in IZ2, while h icEEG 2 showed the inverse pattern. This corroborates previous observations of inter-modal connectivity discrepancies in pathological cortices, while providing the first direct invasive and simultaneous comparison in humans. We also studied time-resolved FC variability multimodally for the first time, finding that IZ1 shows both elevated internal h BOLD 2 and less rich dynamical variability, suggesting that its chronic role in epileptogenesis may be linked to greater homogeneity in self-sustaining pathological oscillatory states.