Two studies tested the hypothesis that the right hemisphere engages in relatively coarse semantic coding that aids high-level language tasks such as joke comprehension. Scalp-recorded event-related brain potentials (ERPs) were collected as healthy adults read probe words (CRAZY) preceded either by jokes or nonfunny controls (“Everyone had so much fun jumping into the swimming pool, we decided to put in a littlewater/platform”). Probes were related to the meaning of the jokes, but not the controls. In Experiment 1a, with central presentation, probes following jokes (related) elicited less negative ERPs 300–700 msec postonset (N400) than did probes following nonfunny controls (unrelated). This finding suggests related probes were primed by the jokes. In addition, unrelated probes elicited a larger anterior positivity 700– 900 msec than did related, as irrelevant stimuli impacted control processes invoked by task demands. In Experiment 1b, probes (CRAZY) were preceded only by sentence-final words from jokes (water) or controls (platform). No ERP effects were observed in Experiment 1b, suggesting the N400 priming effect and the anterior positivity observed in Experiment 1a reflect semantic activations at the discourse level. To assess hemispheric differences in semantic activations, in Experiment 2, ERPs were recorded as participants read probe words presented in their left and right visual fields (LVF and RVF, respectively). Probes elicited a smaller N400 component when preceded by jokes than controls. This N400 priming effect was larger with presentation to the LVF, suggesting joke-relevant information was more active in the right hemisphere. The anterior positivity was observed with RVF but not LVF presentation, suggesting an important role for the left hemisphere in controlled retrieval in language comprehension.