Validity coefficients of cognitive ability tests for predicting job pe
rformance are impressive. Nevertheless, a need exists to supplement co
gnitive ability rests with additional predictors to maximize validity
and utility and to minimize adverse impact-even for tests that are unb
iased according to psychometric criteria. An agenda for research and p
ractice is presented that includes building on the existing validity o
f cognitive ability tests for job selection by expanding both the set
of constructs used to predict job performance and the set of performan
ce criteria used to quantify job performance. Even greater benefits ar
e likely to be associated with advances in knowledge about effective t
raining. Such advances will require substantial investment in well-des
igned, large-scale studies and facile adaptation of lessons from other
domains, such as cognitive psychology, to the organizational-training
context.