Nuclear grade is considered a valuable prognostic factor in mammary ca
rcinomas. Since the histological diagnosis of most of these tumors is
made by ''non expert'' pathologists, it was considered interesting to
find out the reproducibility of general pathologists to define the nuc
lear grade. in order to do this, a series of 15 mammary carcinomas, 10
of them randomly selected and 5 because they were considered difficul
t to classify for nuclear grade, were examined separately by 10 genera
l pathologists. In a first round of observation, each one of them grad
ed the cases according to their own criteria as used routinely, and fo
r a second round they followed a written guide. An analysis of varianc
e was applied to the data and no significant differences were found be
tween observers, neither in the randomly selected cases nor in the tot
al series. The written guide, surprisingly, instead of lowering the di
fferences, increased them. Analysis of the individual performance of o
bservers showed two of them having a great variation between both roun
ds of observation, and this was considered to influence the results of
the whole group. Interobserver performance to discriminate high grade
tumors (G3) from the rest, showed a good correlation in all the parti
cipants. These results allow us to conclude that in this series, exami
ned by general pathologists, an acceptable reproducibility was observe
d, specially when high risk tumors were being identified.