There is accumulating evidence that highly anxious individuals selectively attend to threatening information; however, contrary to expectations, there is no evidence of enhanced processing of threat stimuli in those individuals. We investigated this question by using a sample of 20 University students who were split into two groups consisting of 10 high-anxious and 10 low-anxious subjects according to Spielberger Trait Anxiety Inventory score (median STAI = 40). Without emphasizing speed, subjects were required to decide whether visually presented words (positive, neutral, or threat) matched semantically with previous priming sentences (threat or positive) displayed on a computer screen (altogether, two types of priming sentences and three types of probe words). There was a fixed interval of 1.1 seconds between priming sentences (SI) and probe words (S2) as well as between each priming sentence word. Response time and visual event-related potentials (ERPs) were recorded in both conditions. The results showed that, compared to the Low-Anxious group, the amplitude of N100 and P400 were enhanced for the High-Anxious group in the threat priming conditions. Low-anxious individuals showed almost identical processing of threat-related situations and positive situations. Furthermore, the P400 peak latency was shorter for emotional incongruous probes in high-anxious individuals. ERPs results of the experiment suggest that highly anxious individuals deploy more processing resources to threatening information. This bias in information processing occurs in the absence of any behavioral changes (indicated by reaction times). Attentional bias in anxiety therefore implies that threatening information is given a priority over other information and is more persistently activated in anxiety states.