The overall aim of this research was to investigate the affective characteristics of achieving and underachieving gifted children. In particular, this project examined the academic self-concept, self-expectations for future achievement and academic locus of control of achieving gifted, underachieving gifted and average achieving children. Subjects were chosen from a total nonreferred Form 1 population of 1,220 children from Palmerston North and Feilding intermediate schools. The group intelligence test, the Test of Scholastic Abilities (Intermediate B)(T0SCA) was used as an initial screening device and those students who had a predicted Weschler Intelligence Scale for Children (Revised)(WISC-R) IQ score of 118 or more were administered the Full WISC-R test. Forty one students had a WISC-R Full Score (FS) IQ>125 and were classified as gifted. Seven of these students were classifed into an underachieving gifted group as their actual Performance Achievement Test (PAT) measures were one standard error of estimate or more below their predicted scores as determined by the regression equation method, whereas the remaining 34 were placed into an achieving gifted group. A third group, classified as average achievers, was composed of children who had WISC-R FS IQs ranging between 90 to 110 and whose achievement on the PAT was within one standard error of estimate of prediction. Of the gifted groups, 26 of 34 of the achieving and five of seven of the underachievers were males. Most of the gifted population came from the professional and managerial socio-economic classes. In the average achieving group there were more females (22 of 39) and the full range of socio-economic groups were represented. The Student's Perception of Ability Scale (SPAS) was administered to all three groups to test the hypotheses that significant differences in academic self-concept would be found between all three groups and that over time there would be a perceptible decrease in this self-concept. There was a significant difference in academic self-concept between the achieving gifted and average achieving groups (p<0.05), but no significant differences between underachieving gifted and either achieving groups was found. No significant decline in academic self-concept occurred over time. The self-expectations for future achievement were assessed using the Projected Academic Performance Scale (PAPS) and the hypothesis that there would be significant differences between achieving gifted and underachieving gifted, between achieving gifted and average achievers but not between underachieving gifted and the average group was confirmed (p<0.01). The self-expectations also declined as predicted over time for all three groups with the greatest difference noted for the achieving gifted children. The final construct, the academic locus of control, was measured using the Intellectual Achievement Responsibility Questionnaire (IAR). No significant differences were found between the three groups. All subjects ascribed responsibility for success to internal causes and were more external for failure outcomes. This result was contrary to the hypotheses that achieving gifted children would be more internal for both success and failure outcomes and that the underachieving gifted would be more external on failure outcomes than either achieving group. Discriminant function analysis showed that 71.9 percent of cases were correctly identified (hits) and this appeared to justify at least the use of the self-expectations for future achievement construct in discriminating achieving gifted from underachieving gifted children. On the basis of the above findings it was concluded that achieving and underachieving gifted children did not differ greatly in their school-related affective characteristics. Academic self-concept and locus of control did not discriminate between these two groups. In fact, the locus of control results suggested the need for the use of an alternative instrument. Self-expectations for future achievement were, however, significantly different for these two groups and this variable was considered to be the most useful for further investigation and the most likely target for the remediation of underachievement.