Behavioral studies suggest that argument phrases are processed more easily than adjunct phrases (Clifton, Speers, & Abney, 1991; Kennison, 1999, 2000; Schultze & Gibson, 1999; Speer & Clifton, 1998). Here we investigated the influence of argument structure on lexical integration by comparing the comprehension of argument and adjunct phrases in simple active sentences in an event-related potential (ERP) experiment with 64 channels. We monitored the latency and amplitude of the N400 component, a marker of semantic integration (e.g. Kutas & Federmeier, 2000), at the centro-parietal electrodes between 300 and 600 ms, measured for arguments and adjuncts to that obtained in the odd condition. Subjects listened to naturally produced sentences that ended with either a plausible argument, a plausible adjunct, or a semantically unrelated word (baseline condition). The results showed that the N400 was more reduced for the argument and adjunct compared to the unrelated condition, and most interestingly that the N400 in the argument condition was more reduced than that in the adjunct condition as early as 300 – 400 ms. The results support the hypothesis that argument phrases are easier to integrate than adjunct phrases in sentence processing and have implications for current models of lexical integration and syntactic parsing.