To control a discrete manufacturing system (DMS), one must first estimate i
ts current transient state. If control is to occur in a closed-loop mode, t
hen a method is required to estimate the system state automatically. Howeve
r, the state of a DMS changes rapidly, and the estimation of its transient
state poses subtle challenges. This paper evaluates four methods for transi
ent state estimation of a DMS, including a fuzzy algorithm that estimates t
he qualitative trajectory of system congestion. None of the methods require
that any restrictive assumptions be made that could limit their applicabil
ity to practical systems. The effectiveness of the four methods for transie
nt state estimation is demonstrated by their inclusion in an application to
on-line, closed-loop control of order lead time for a job shop. Simulation
experiments show that most of the methods for transient state estimation a
re fast and accurate enough to significantly improve the performance of the
controller for order lead time. Following detailed analysis of the results
, the paper concludes with a discussion of the intuitions gained from this
research regarding the nature of transient states in a DMS.