Predicting the fertilizer requirements of an oil palm plantation has long been a difficult task. Two main methods have emerged. Leaf analyses (LA) were used for fertilization management as early as the 1950s. Leaf contents are compared to optimum references, making it possible to adjust the fertilizer rates applied in each block. Another approach, based on the nutrient balance (NB), is to evaluate and replace nutrients that are exported from the field, or immobilized by the plant. Plantations must adopt environmentally friendly practices; in particular, fertilizer inputs must be estimated with sufficient precision to achieve the highest possible yields, without applying excessive amounts of nutrients in relation to plant demand and the storage capacity of soils. We questioned the relevance of each method for achieving these objectives. We did so using some long-term fertilization trials to compare the optimum N and K rates recommended by each method in the adult phase. It appeared that LA led to moderate rates compared to NB. It also appeared that calculating a precise nutrient balance on a field scale was hampered by a lack of precise information (i) about the biomasses produced and their composition and (ii) about the highly variable outputs of the environmental losses. On the other hand, LA provided a simple indicator of the ability for each block to achieve its potential yield. We believe that this perfectible method is more protective of the environment, without the risk of a significant decrease in yields or a decrease in soil mineral reserves.