This talk contrasts the drug testing methods of two sixteenth-century alchemical empirics. Andreas Berthold validated his Paracelsian poison antidote (also deemed a cure-all) by letting learned physicians conduct poison trials at German courts, in which test subjects (several dogs and a convict) took poison, followed by the antidote. Georg Amwald, in contrast, scoffed at this method and instead included patient testimonial letters as evidence of the efficacy of his panacea poison antidote - a method also used by earlier alchemists such as Leonardo Fioravanti but derided by physicians. These cases elucidate the tricky problem of proof and evidence in early modern drug testing. While poison trials were used at princely courts all over Europe and appeared to give a definitive answer, they could also be dismissed as singular tricks. Testimonial letters, meanwhile, had perceived problems of trustworthiness. I argue that the boundaries of proof were contested and depended largely on the professional designation of the tester.