When polling systems are used to model real-world systems, it is typic
ally assumed that the server switches continuously (''roves'') even wh
en there are no waiting jobs in the system. However, requiring the ser
ver to be patient, instead of having it rove, might be more realistic.
Furthermore, operational control of these systems can be improved by
knowing answers to questions like ''under what circumstances should a
roving server be patient?'' and ''at which stations?'' This paper anal
yzes the patient server model and provides explicit expressions for th
e waiting time distributions, the mean waiting times and the pseudo-co
nservation law. Several variants of the patient server model are consi
dered. We show that while the patient server mechanism is generally be
tter than the roving server mechanism in the work-in-process (WIP) red
uction sense, there do exist cases where roving is better. Counter-int
uitive examples where reducing switchover time can increase WIP are al
so reported.