A number of indices are currently used by ecologists to measure the si
milarity or difference between community samples. Nearly all are symme
trical, in that the degree df similarity is equal for both parties bei
ng compared. There are situations, however, where an asymmetrical meas
ure is more appropriate, most notably in studies of dispersal and its
effects. I propose 2 variants of a potentially useful asymmetrical sim
ilarity index for use in such circumstances. In most respects, they ar
e well behaved. reflecting the similarity and nestedness of samples, w
hile being relatively insensitive to sample size and species diversity
. They are also simple to compute and intuitively meaningful. Nonethel
ess, they provide anomalous results in cases where one plot contains a
nested subset of the commonest species of another. I welcome suggesti
ons for a better asymmetric index of similarity.