In 1992, a statistical analysis of data from 61 studies of semen quali
ty among normal men led to the conclusion that human sperm count fell
of 40% from 1940 to 1990. This meta-analysis has since been invalidate
d for methodological reasons as well as for statistical reasons, but i
t caused enormous concern to both the scientific community and to the
international media. A decline of human fertility was speculated. To d
ate, the most popular hypothesis offered to explain this alleged decli
ne has been increasing exposure to environmental estrogen mimicking ch
emicals. However, there is no evidence that male or mammal fertility i
s declining. Moreover, the sperm count of breeding mammals did not dec
lined in the meanwhile. Since 1992, numerous papers reported on men in
vestigated during the last 20 years have shown conflicting results, fr
om sperm count improvement to sperm count decline. However, several pu
blications included methodological and analytical biases. In fact, the
techniques used for semen analysis have to be questioned. It is a sub
jective exam, lacking laboratory standards and quality control procedu
res. This induces very important variations between laboratories and b
etween biologists. For the sperm count itself numerous errors can occu
r, provoked by the technique,the equipment and the reader: for the sam
e sperm, the coefficient of Variation can exceed 40% between two techn
icians. Therefore, the current techniques of semen analysis cannot war
rant epidemiological studies. It is indispensable to admit the limits
of the semen analysis in order to improve as much as possible its qual
ity and its reliability.