We introduce the log-normal distribution as a simple parametrization of multiplicities in hadron-hadron minimum-bias scattering and show that it fits data at all available energies, including the highest energy (pp̄ at √s=900 GeV) were the negative binomial distribution cannot describe the data well. The log-normal distribution is a two-parameter statistical probability function that has a simple and appealing physical interpretation. In statistics theory it arises in the case of multi-step cascade processes where each step multiplies the previous outcome. This gives a natural connection to state-of-the-art models for multiparticle production involving QCD parton cascades and hadronization.