The human brain has evolved to process motivationally relevant information in an optimized manner. The perceptual benefit for emotionally arousing material, termed motivated attention, is indexed by electrocortical amplification at various levels of stimulus analysis. An outstanding issue, particularly on a neuronal level, refers to whether and how perceptual enhancement for arousing signals translates into modified processing of information presented in temporal or spatial proximity to the affective cue. The present studies aimed to examine facilitation and interference effects of task-irrelevant emotional pictures on subsequent word identification. In the context of forced-choice lexical decision tasks, pictures varying in hedonic valence and emotional arousal preceded word/ pseudoword targets. Across measures and experiments, high-arousing compared to low-arousing pictures were associated with impaired processing of word targets. Arousing pleasant and unpleasant pictures prolonged word reaction times irrespective of stimulus-onset asynchrony (80 msec, 200 msec, 440 msec) and salient semantic category differences (e.g., erotica vs. mutilation pictures). On a neuronal level, interference was reflected in reduced N1 responses (204–264 msec) to both target types. Paralleling behavioral effects, suppression of the late positivity (404–704 msec) was more pronounced for word compared to pseudoword targets. Regional source modeling indicated that early reduction effects originated from inhibited cortical activity in posterior areas of the left inferior temporal cortex associated with orthographic processing. Modeling of later reduction effects argues for interference in distributed semantic networks comprising left anterior temporal and parietal sources. Thus, affective processing interferes with subsequent lexico-semantic analysis along the ventral stream.