Placebos can reduce pain by inducing beliefs in the effectiveness of an actually inert treatment. Such top-down effects on pain typically engage lateral and medial prefrontal regions, the insula, somatosensory cortex, as well as the thalamus and brainstem during pain anticipation or perception. Considering the level of large-scale brain networks, these regions spatially align with fronto-parietal/executive control, salience, and sensory-motor networks, but it is unclear if and how placebos alter interactions between them during rest. Here, we investigated how placebo analgesia affected intrinsic network coupling. Ninety-nine human participants were randomly assigned to a placebo or control group and underwent resting-state fMRI after pain processing. Results revealed inverse coupling between two resting-state networks in placebo but not control participants. Specifically, networks comprised the bilateral somatosensory cortex and posterior insula, as well as the brainstem, thalamus, striatal regions, dorsal and rostral anterior cingulate cortex, and the anterior insula, respectively. Across participants, more negative between-network coupling was associated with lower individual pain intensity as assessed during a preceding pain task, and there was no significant relation with expectations of medication effectiveness in the placebo group. Altogether, these findings provide initial evidence that placebo analgesia affects the intrinsic communication between large-scale brain networks, even in the absence of pain. We suggest a theoretical model where placebo analgesia might affect processing within a descending pain-modulatory network, potentially segregating it from somatosensory regions that may code for painful experiences.