To serve as a useful guide to fertilizer management, a nutrient diagnostic method should identify cases in which a nutrient limits yield, as well as cases in which the nutrient does not limit yield. Erroneous diagnoses can lead either to unnecessary fertilizer application or failure to apply fertilizer when a yield response would result. We propose a process of prescient diagnostic analysis to evaluate the utility of diagnostic methods based upon comparing individual diagnoses to corresponding responses to fertilizer applications. A nutrient diagnostic method is evaluated upon the incidence and yield consequences of its true and false positive and negative diagnoses. Evaluation criteria integrate these effects to facilitate comparison of various diagnostic systems. We propose a protocol for evaluating diagnostic methods based upon this reasoning.