Our ability to recognize tens of thousands of words is attributed to a rich lexicon in our brains, referred to as the orthographic lexicon. Understanding how this lexicon is organized in the brain is key to uncovering the neural mechanisms of visual word recognition. Event-related potentials (ERPs) are an effective tool for investigating these mechanisms. A widely observed phenomenon in orthographic processing is a 250-ms ERP difference between real words and pseudowords (or between high-frequency (HF) and low-frequency (LF) words). However, this 250-ms ERP difference has been explained by competing theories—one attributing it to the activation of the orthographic lexicon, and the other suggesting it represents a prediction error signal related to orthographic processing. To resolve this debate, the present study uses handwritten words as control stimuli alongside printed words, as the 250-ms ERP difference is not observed with handwritten words. ERPs were obtained during an implicit reading task (color decision) for printed HF words, printed LF words, handwritten HF words, and handwritten LF words. The results show that the 250-ms ERP difference is significant when comparing printed LF words to handwritten LF words, but not when comparing printed HF words to handwritten HF words. This finding suggests that the 250-ms ERP difference reflects a prediction error rather than activation of the orthographic lexicon. These results support the Interactive Account of orthographic processing, clarify previous ERP findings in the literature, and underscore the potential applications of the 250-ms ERP difference (labeled as ND250) in future research.