Experimental mechanical tests on model-scale prototype tidal turbine blades are intended to characterize the strength and adequacy of the blade scantling. Whereas static mechanical tests simulate the extreme static load expected to be encountered during the service life, the main challenge behind fatigue tests is to recreate a loading condition that is representative of the entire service life. Simulating environmental loads for industrial applications generally involves statistical methods that make use of a large database of site measurement data. Such classical methodologies result in load spectra which are statistical representations of the realistic conditions. The unique selling proposition (USP) of a tidal environment is its predictability - an accurate forecast is obtained from planetary orbital motions. Within this paper, an alternative approach is proposed which aims to exploit the predictable nature of the tidal environment to generate a realistic fatigue load spectrum, in deviation with the classical methods based on statistical distributions. Thereafter, a model scale extrapolation of the full-scale conditions is performed, with the law of similitude based on the stresses experienced at the ply level. Ply-by-ply fatigue stress spectra are generated to estimate the fatigue life of the blade, using uni-directional S-N curves.