We have considered German domestic and international football score data with respect
to certain phenomenological probability distributions as well as microscopically motivated
models. The Poisson distribution resulting from the assumption of independent scoring probabilities
for the opposing teams does not provide a satisfactory fit to any of our data. Many
data sets are rather well described by the negative binomial distribution considered before
[3], however, some cases have heavier tails than can be accommodated by this distribution
and, instead, rather follow a distribution from extreme value statistics