This field study develops training for fault-finding in a hot strip st
eel mill and evaluates its effectiveness. A prescriptive model of faul
t-finding was developed to overcome the errors made by technical perso
nnel whilst fault-finding. This model breaks down fault-finding into t
wo main goals: initial symptom identification and narrowing dawn the l
ocation of the fault to one sub-system. A computer-based training syst
em was designed to train achievement of these goals by inculcating app
ropriate knowledge and by providing practice at simulated faults. The
simulations had high physical and psychological fidelity with informat
ion available in the mill. The pre- and post-training performance was
compared for eight technical personnel (four experienced and four appr
entices) who tackled four matched faults. There were improvements afte
r training in achievement of both goals. This was confirmed by qualita
tive analyses of the concurrent verbal reports obtained from the four
apprentices whilst fault-finding.