A recently identified dinoflagellate, Pfiesteria piscicida, has been implic
ated as a cause of fishkills in mid-Atlantic estuaries. To date, field evid
ence sup porting this argument has consisted of samples, analyzed for the p
resence of the toxic Pfiesteria forms, gathered during a fishkill. I presen
t a probabilistic approach to examine the use of this kind of a posteriori
information as an indication of cause and effect relationships. The analysi
s shows that the conditional probability of the presence of Pfiesteria afte
r a fishkill has begun provides little support for Pfiesteria as a cause of
fishkills, without also knowing the probability of Pfiesieria's presence u
nder all conditions. Documenting the relative presence of toxic life stages
during fishkills and under non-fishkill conditions will provide supporting
evidence to assess Pfiesteria's role in fishkills. However, proving that P
fiesteria causes estuarine fishkills using only 'after the fact' informatio
n is essentially impossible.