Anxious subjects present attentional disorders that are manifest with an increased bias towards threatening contents stimuli. In tasks derived from the Stroop task (such as emotional Stroop, a variant of the classic Stroop task) congruence between anxious themes or manifestations and stimuli content induces information processing changes leading to a slowness of response speed. In this case, results are similar to those obtained in signal detection tasks either when information is visually or auditorily presented. In anxious subjects an inconscious activation provoked by anxiogenic words is observed. Because such activation is independent from the semantic content of the words, an emotional priming has been hypothesized. Berck formulated an hypervigilance theory according to which anxiety provokes a selective distractibility regarding non pertinent stimuli. Such attentional selectivity would be responsible of a cognitive vulnerability in anxious subjects. State but not trait anxiety induces working memory performances deficit. On the bases of Baddeley's working memory framework, Eysenck proposed that anxiety uses part of the limited attentional capacity, placing the subject in a dual task situation. In that, he has to cope with pertinent information and anxiety generated information. If anxiety leads to better performance in simple tasks by recruiting motivational capacities, in tasks with high information content, anxious subjects performances are impaired. Changes in the long-term memory do not seem to fit with the theoretical models based on cognitive impairment observed in patients suffering from depressive states. Anxious subjects presented a memory bias towards anxiogenic information in implicit memory tasks. But experimental data are still too searce to describe implicit performance of anxious subjects and more systematic studies are therefore needed.