To highlight the impact of peak height analysis on the inference to be drawn from typical, two persons mixtures (2PMs), a number of real high-quality – either "balanced" or "unbalanced" – 2PMs were subjected to a classical statistical analysis based on the likelihood of a known contributor to be therein. We calculated a LR value based on the genetic markers belonging to a real contributor (S1) and to a number of false but compatible suspects using a discrete/semi-continuous method (as implemented in a shareware of widespread adoption[LR-mix Studio]) and a quantitative approach of our own conception (built into a Microsoft Excel spreadsheet). Quantitative information conferred by the height of the peaks was irrelevant in the case of "balanced" mixtures but played a key role in eliciting correct inference from "unbalanced" 2PMs.