The goal of this study was to test whether semantic priming in response to a word that immediately follows a related sentence or single-word is age-invariant. This was done using the N400 event-related brain potential (ERP). N400 amplitude varies inversely with semantic priming (Kutas, 1980), and this effect is larger for sentence primes than for word primes (Kutas, 1993). ERPs were recorded while 20 older ( M = 71.5, SD = 6.4) and 20 young ( M = 23.0, SD = 2.3) subjects read word-pairs and highly constrained sentences. Each final word varied on the degree of relatedness to the preceding context, with some being highly related (BC), moderately related (R), or unrelated (U). An Age X Context X Relatedness repeated-measures analysis of variance (ANOVA) was conducted on N400 amplitude at the .05 level of significance. Significant results were further analyzed with Tukey A post-hocs also conducted at the .05 level of significance. In young subjects, N400 amplitude showed the expected N400 effect gradient (U > R > BC) in both the sentence and word-pair contexts. N400 responses in older subjects showed no discrimination between the conditions (U = R = BC) for the word-pairs and limited discrimination (U > BC) for the sentence endings, indicating that older subjects were able to benefit from the constraints of a sentence context. These results provide electrophysiological evidence of age-related changes in semantic priming, and are interpreted in the context of the inhibition-deficit hypothesis (Hasher & Zacks, 1988) whereby older adults fail to inhibit related but irrelevant items in working memory.