Startle eyeblink modification was examined as a measure of allocation of attentional resources during active attention tasks in the early stage of information processing. Fifty-five participants were presented with a series of 250- and 40-ms tones of either high or low pitch which were followed by startle-eliciting stimuli at a lead interval of 120 ms. Attentional allocation was manipulated by instructing one group (Passive) to simply listen to the tones; the second group (Active 1) to count the number of low tones and the third group (Active 2) to count the long high-pitched tones and the short low-pitched tones. Startle eyeblink was significantly more inhibited for the Active 1 group than the Passive group (control) with no significant difference between the two directed attentional conditions (Active 1 and Active 2 groups). However, across the three attentional groups, the degree of startle eyeblink modulation appeared to reflect the degree of attention allocated to the task. The results support the utility of the startle probe in evaluating controlled attentional allocation during the early stages of information processing.