The purpose of this study was to compare two competing models as an explanation of the relationship between intelligence and sustained attention in educationally at-risk kindergarten children. One model assumes that lower-IQ subjects allocate greater amounts of attentional resource of information-processing tasks than higher-IQ subjects, whereas the other model assumes that a "less-than" optimal level of arousal is associated with decrements in task performance across time. Twenty-nine teacher-nominated at-risk and 29 normal achieving kindergarten students were administered the Wechsler Preschool and Primary Scale of Intelligence-Revised (WPPSI-R) and vigilance taks. Signal detection measures of stimulus detectability (d'), decision criterion (beta), correct detections, and false alarms were used to assess children's sustained attention across three time periods (2, 4, and 6 min). The important results were (a) high-risk children were inferior on d' measures when compared to normal achieving children, (b) vigilance measures did not vary over time in either group, and (c) intelligence and vigilance shared a common factor in high-risk, but not low-risk, children. The results suggest that children educationally at risk suffer deficits related to attentional capacity for processing information.