We have investigated possible models for the evolutionary status of the bla
ck hole soft X-ray transient GRO J1655-40. A secondary star evolving across
the Hertzsprung gap and filling its Roche lobe can be ruled out, because t
he expected mass transfer rate is over 1000 times greater than the estimate
d upper limit based on observations. A secondary star evolving across the H
ertzsprung gap, not yet filling its Roche lobe, but transferring mass via a
n enhanced stellar wind cannot be ruled out, but it may be unsatisfactory,
because the deep convective envelope, normally associated with slow red-gia
nt winds, has not really developed at the observed temperature. We propose
that the secondary of GRO J1655-40 may be a main-sequence star in which con
vective overshooting, or some equivalent mixing, occurs at the core-envelop
e boundary. With such a model, the predicted mass transfer rate can be as l
ow as 16 times the observational estimate, and we indicate how this factor
might be accounted for so as to fully reconcile this model with the observa
tions.