We show some inaccuracies in recent arguments claiming that in the lon
g-time limit the initially correlated A + B --> 0 diffusion-limited re
action can be faster than the A + A --> 0 one. With the errors correct
ed, these arguments seem to confirm the former theory of Toussant and
Wilczek according to which the global rate of both reactions should ul
timately be the same. This hypothesis is also supported by our numeric
al simulations of a two-dimensional lattice system.