A recent paper by Hartle [Phys. Rev. D 51, 1800 (1995)] proposes a def
inition of ''spacetime information'' - the information available about
a quantum system's boundary conditions in the various sets of decoher
ing histories it may display - and investigates its properties. We not
e here that Hartle's analysis contains errors which invalidate several
of the conclusions. In particular, the proof that the proposed defini
tion agrees with the standard definition for ordinary quantum mechanic
s is invalid, the evaluations of the spacetime information for time-ne
utral generalized quantum theories and for generalized quantum theorie
s with nonunitary evolution are incorrect, and the argument that space
time information is conserved on spacelike surfaces in these last theo
ries is erroneous. We show, however, that the proposed definition does
, in fact, agree with the standard definition for ordinary quantum mec
hanics. Hartle's definition relies on choosing, case by case, a class
of fine-grained consistent sets of histories. We supply a possible gen
eral definition of a class of sets that includes all the sets consider
ed in Hartle's paper and that generalizes to other cases.