Fruit splitting is a preharvest physiological disorder that occurs in many commercially important fruit species, including some citrus cultivars such as Navel oranges, Valencia oranges, and some mandarins. The phenomenon is affected by both genetic background and environmental conditions, causing heavy fruit loss in splitting-prone cultivars in some years. For instance, high levels of irrigation usually enhance splitting incidence. The phenomenon initiates at the end of the summer, toward fruit maturation, and to date, there is no way to predict splitting incidence. Here, we tested the ability of nondestructive magnetic resonance imaging (MRI), which does not alter the dimensions of the internal fruit tissues, to predict splitting incidence in 'Nova' mandarin fruit populations of trees subjected to two levels of irrigation (low and high). Samples, collected about two months prior to splitting appearance, were sorted by their internal tissues dimensions. Among all measured tissues, sorted navel dimensions showed the best correlation with splitting incidence. This was determined by dividing the MRI-sampled fruits into two populations, according to the actual splitting incidence, as calculated at the end of the season. Prediction of the splitting percentage in the low irrigation fruit population was than possible in the high irrigation fruit population, and vice versa. These results demonstrated the power of MRI to predict splitting probability as early as 2 months before split fruit are found. In addition to its potential practical application, the ability to predict splitting probability in a given fruit population could help elucidate the mechanism underlying the disorder.