We investigate the possibility that alpha(s) freezes as function of N-
f within perturbation theory. We use two approaches - direct search fo
r a zero in the effective-charge (ECH) beta function, and the Banks-Za
ks (BZ) expansion. We emphasize the fundamental difference between qua
ntities with space-like vs, those with time-like momentum. We show tha
t within the ECH approach several space-like quantities exhibit simila
r behavior. In general the three-loop ECH beta functions can lead to f
reezing for N-f greater than or similar to 5, but higher-order calcula
tions are essential for a conclusive answer. The BZ expansion behaves
differently for different observables. Assuming that the existence of
a fixed point requires convergence of the BZ expansion for any observa
ble, we can be pretty sure that there is no fixed point for N-f less t
han or similar to 12. The consequences of the Crewther relation concer
ning perturbative freezing are analyzed. We also emphasize that time-l
ike quantities have a consistent infrared limit only when the correspo
nding space-like effective charge has one. We show that perturbative f
reezing can lead to an analyticity structure in the complex momentum-s
quared plane that is consistent with causality. (C) 1998 Elsevier Scie
nce B.V.