The concept of evolutionary constraint is central to current perceptio
ns of the evolutionary process, but operationally, it is difficult to
apply. Much of this difficulty stems from conceptual ambiguity and inc
onsistent usage. The broadest conception of constraint as ''historical
contingency'' has little pragmatic value because it universalizes con
straint to a property of life. Likewise, equation of stabilizing selec
tion with constraint creates unnecessary conceptual redundancy. Concep
tions of constraint that emphasize mechanism over phenotypic or phylog
enetic patterns tend to eliminate redundancy and restrict constraint t
o a force that shapes the action of natural selection, and which may o
ppose it. Constraint is a property of characters, not lineages, and at
this level is always negative in the sense of limitation. However, ch
aracter constraint is neutral to organismal adaptation and, therefore,
can have either negative or positive evolutionary effects at the line
age level (i.e., hamper or promote organismal adaptation). Constraint
hypotheses can be framed from either a posteriori or a priori perspect
ives that show that constraint is sensible only when bounded within a
relative time-frame. Both stasis and parallelism have been invoked as
phylogenetic manifestations of constraint and these alternate concepti
ons can lead to opposing hypotheses about the time and place of action
of constraint. Disallowing pattern as prima fade evidence of constrai
nt avoids this conflict. Homology and Bauplan might reflect the action
of constraint at the level of one and many characters, respectively,
but the failure to evolve at any given hierarchical level should not b
e taken as direct evidence of constraint.