Several forensic disciplines that sent people to prison for decades turned out to have almost no scientific foundation, and the legal system admitted them anyway for a specific and unflattering reason. This article explains how forensic evidence actually gets into an American courtroom. It sorts the disciplines honestly into the ones with a real quantitative foundation, the ones that have collapsed, and the ones that are genuine but were badly oversold. It covers the gap between what a method can support and what an examiner says on the stand, which is where cases are actually won. It works through what DNA can and cannot establish once mixtures, transfer, and proprietary software are involved, the confrontation problem of who must testify about a laboratory report, the disclosure obligations laboratories routinely fail, and the cognitive bias that operates without anyone lying. It ends with what actually works.