In adults, short-lead prepulse inhibition and long-lead prepulse facilitation of startle are greater during attended than ignored prestimuli. To examine these phenomena in children, fourteen 9- to 12-year-old boys completed a tone discrimination task in two sessions separated by 1 week. During each tone series, participants attended to one pitch and ignored the other. Startle probes (102 dB) were presented 120, 240, 2,000 or 4,500 ms following the onset of two-thirds of tones, and during one-third of intertrial intervals. Eyeblink EMG startle was recorded. Percent prepulse inhibition at 120 ms was greater for attended than ignored stimuli in Session 1 but not Session 2. Long-lead prepulse facilitation was marginally greater for attended than ignored tones and did not vary across sessions. Test–retest reliability was moderate during attended prestimuli but was modest during ignored prestimuli. Reliability of attentional modification was poorest at 120 ms and strongest at 4,500 ms. Overall, this study extended prior work in adults and provided a basis for further study of controlled attentional modification of startle in children.