[Truncated abstract] The overall aim of the research reported in this thesis was to develop and validate an instrument that would identify young children at educational risk. To achieve this, four separate but interrelated studies were conducted. Study One sought to explore the construct of young children at educational risk from the perspective of school psychologists and teachers through a series of open ended interviews. To this end, four school psychologists in charge of centres for severely disruptive children, and seven psychologists and three teachers involved in alternative education programs for alienated youth were interviewed. The findings of Study One revealed a broad range of factors and issues that contribute to a child being placed at educational risk. Interviewees specifically highlighted the significance of behavioural, learning, social, psychological and family factors which contributed to children's at educational risk status. The findings of Study One were incorporated into Study Two, the overall aim of which was to develop and validate a new instrument for the early identification of children at educational risk. Initially, 152 items were generated from the literature critically reviewed in Chapter Two, existing instrumentation and the Study One interview findings. ... In Study Four the construct validity of the CaERI was investigated by examining whether scores on the instrument differentiated between at educational risk groups of children and a group of not at educational risk children. The total data were first investigated for normality of distribution and returned a Shapiro-Wilk statistic of .99 and a significance level of .069 indicating that the combined data set were normally distributed. The individual CaERI scores for the three individual domains were then investigated and Shapiro-Wilk scores ranging from .96 to .99 and significance levels of .13 to .84, supported the assumption that the data are normally distributed. Significant between groups differences were found on the Behaviour Domain, where LDC children scored significantly lower than the regular class at educational risk and SPER-C children. The regular class at educational risk children also scored significantly lower than the SPER-C children which suggest that in the Behaviour Domain the instrument is sensitive enough to differentiate between the three groups. Significant between group differences were also evident between the LDC and regular class at educational risk and the SPER-C groups, with the LDC children scoring lower than the other two groups in the Social and Psychological Domains. Although the SPER-C group children scored higher than the regular class at educational risk group the difference was not significant. Similarly, a gender comparison of the at educational risk groups showed significant differences between males and females on the Behaviour Domain, however, the differences were not significant on the Social or Psychological Domains. A comparison of 20 not at educational risk students with a matched sample of 20 regular class at educational risk students found significant differences between the v groups on all domains. All findings are discussed and interpreted in line with the current research literature and are used to make suggestions for further research.