The results of two carefully instrumented horizontal loading pile tests in clay are compared with the predictions obtained by existing methods for deriving the P-y curves from DMT data. The piles were "Multiton" mandrel-driven cast-in-situ piles, 57 m long, 0.50 m in diameter. The soil, in its upper part - relevant to the horizontal loading - is soft clay, with a 4 m desiccation crust. The first test pile was loaded with the head free to rotate, the second with the head restrained from rotation. The maximum applied horizontal loads for the two piles were 320 and 420 kN respectively, with corresponding horizontal displacements of approximately 110 mm and 30 mm. The Robertson et al. (1989) method predictions are so close to the observed results that, considering also previous validations, one may be tempted to conclude that this method has solved for good the problem of the linkage between P-y curves and DMT data (for "ordinary" clay, soft to moderately stiff, under static monotonic short-term one-way loading). Of course further validations can only be encouraged. A "simplified formulation", providing similarly accurate predictions, is also presented.