When using molecular markers to study genetic variation, either the sampled
individuals can be analysed individually or the individuals can be pooled
and only the pools analysed (pooled samples). A theoretical investigation w
as carried out into the use of pooled samples in the detection of alleles a
nd providing maximum likelihood estimates of allele frequency. The results
show that, in many cases, pooled samples are more efficient than samples of
individuals. Of the different pool sizes studied, small pools containing t
wo or three individuals showed the smallest expected squared error of allel
e frequency estimates.