Shigeo Shingo's 'SMED' methodology has been at the forefront of retrospecti
ve changeover improvement activity since the mid-1980s. The 'SMED' methodol
ogy, which emphasizes that improvement should be sought primarily by rearra
nging changeover elements into external time, has been widely acclaimed and
has been widely assimilated into academic texts and industrial training ma
terial. To date, no known critical evaluation of the methodology has been u
ndertaken. This paper, in which a case study is also presented, argues that
in the sequential application of stages, the 'SMED' methodology (including
the sequential application of improvement techniques that are assigned to
those stages) need not always represent an effective improvement route. The
'SMED' methodology's dominant objective of translating tasks into external
time is also considered. The paper argues that the 'SMED' methodology does
not sufficiently promote some important improvement options, particularly
those that seek to reduce the duration of existing changeover tasks or elim
inate them altogether. Opportunities for improvements of this type particul
arly arise when design changes to the existing manufacturing system are con
templated. The issue of design in the context of incremental kaizen improve
ment is also investigated.