Amplified fragment length polymorphisms (AFLPs) currently are among the mos
t widely used marker systems. In many studies, AFLPs are analyzed on the ba
sis of the presence or absence of a band on an electrophoretic gel. As a re
sult, dominant homozygous individuals are not distinguished from heterozygo
us individuals, resulting in a considerable loss of information. This artic
le shows how codominant information can be obtained if the amount of PCR pr
oducts is quantified. Due to measurement variation, genotyping on the basis
of such information is not error-free. We propose use of normal mixture di
stributions to determine the most likely genotype, given the data. The meth
od is exemplified using AFLP data from sugar beet.