A number of authors have proposed that self-assessment may represent a vali
d tool fur use in assessment situations. Research suggests that those who a
re accurate in self-assessments of performance actually perform more effect
ively, e.g. in the job or at an assessment centre. This study examined the
properties of two measures of self-assessment (absolute accuracy of self-as
sessment and degree of inaccuracy of self-assessment) in a graduate assessm
ent centre. Results showed that an accurate self-rating of performance on o
ne assessment centre exercise reliably distinguished between accepted and r
ejected candidates. Further, and importantly, this effect remained after va
riance attributable to psychometric tests and demographic variables had bee
n controlled fur. That is, some accurate self-assessments appear tl, have i
ncremental validity above and beyond psycho metric tests. However, the magn
itude and direction of self-assessment inaccuracy was significantly differe
nt between the accepted and rejected candidates for three of the four exerc
ises. These results suggested that while variables associated with self-ass
essment may impact on the size of the inaccuracy of self-assessments, the i
mpact of such variables map not lead to accurate self-assessment.