For the first time, anxiety-related attentional bias was examined by considering separately its time-course and the influence of processing time. High- and low-trait-anxious undergraduates performed a color-identification task: Emotional-neutral word pairs were presented for 14 or 500 ms; color appeared at word onset or after 514 ms, at emotional word position or at the other position. Word duration (processing time) and color onset time (moment at which bias was 'tapped') were manipulated independently. Relatively fast responses when emotional word and color spatially correspond reflect attentional bias for that word. High-trait-anxious, compared to low-trait-anxious, undergraduates showed biases for physical-threat words and positive words (second half of the task). These group differences did not significantly vary as a function of duration or SOA. The findings suggest a content-specific bias for threat in non-clinical trait-anxiety, very quickly upon stimulus onset, even when only little time is available for stimulus processing. This bias does not vary over a period of 500 ms, also not when further stimulus processing is possible. It is recommended to further investigate the precise temporal characteristics of anxiety-related attentional biases.