In patent-infringement litigation, if no established royalty for the patent in suit has emerged from multiple market transactions at a readily observable price, then the finder of fact needs to infer a reasonable royalty from the many factors identified in the Georgia-Pacific framework. The well-recognized problem with the Georgia-Pacific framework is that it poses many potentially relevant questions but does not say how the finder of fact should weight the answers. The case law offers no algorithm or decision tree for the finder of fact to follow. Courts find expert testimony inadmissible if it does not apply intellectually rigorous economic methods and principles to the facts and data of the case to produce results that are replicable and falsifiable. With modest effort, and without repudiating existing precedent, the courts can make the Georgia-Pacific framework far more coherent, predictable, and intellectually rigorous. From an economic perspective, that framework ultimately leads the finder of fact, first, to determine the gains from trade—which economists call “surplus”—arising from a hypothetical, voluntary negotiation between a willing licensor and a willing licensee just before the moment of first infringement and, second, to divide that surplus between the licensor and licensee according to their relative bargaining power. For brevity and clarity, I call these two culminating steps the surplus-division principle. This principle is more reliable than purporting to set a reasonable royalty on the basis of a mathematical theory (such as the Nash bargaining solution) that is too abstract to fit the facts and data of the case. It is also more reliable than an expert’s idiosyncratic and nonfalsifiable claim to have balanced the totality of the circumstances in light of his professional experience. In contrast to both a theoretical black box and an expert’s ipse dixit, the surplus-division principle uses elementary principles of microeconomics to give coherence to the Georgia-Pacific factors that courts have already defined and applied. The result enables the finder of fact to determine a licensor’s minimum willingness to accept and a licensee’s maximum willingness to pay for the patented technology, and thereby to define the bargaining range for a hypothetical negotiation. This method is robust across different factual scenarios and multiple defendants.