The conditional probability of failure of a structure, given that it h
as already survived for T years, shows that structural reliability inc
reases for service proven structures. Further, construction loads act
essentially as proof loads, and so provide a means to detect the prese
nce of gross construction errors. The present paper investigates the e
ffect of construction error, proof load tests prior to occupation, rea
listic construction loads, and dead loads and office floor live loads
on the probability of failure of service proven structures. The human
reliability analysis (HRA) approach is used to simulate the effects of
construction error. Results obtained from a Monte-Carlo computer simu
lation analysis showed that up to 50% of failures will occur during co
nstruction for reinforced-concrete beams constructed without engineeri
ng inspections. This suggests that construction loads act essentially
as proof loads; hence, for structures with low live-to-dead load ratio
s the adverse consequences of construction error will most likely beco
me evident during construction. Nevertheless, engineering inspections
is a more effective error control measure.