To achieve the full potential of tidal stream energy, developers are incentivised to use larger blades on tidal turbines. This requires validation of blade structural designs through full-scale blade fatigue tests to de-risk the engineering process. However, the loading scenarios encountered in testing facilities and those in reality could be significantly different, which induces errors in blade loads and fatigue damage. Here we characterise the unsteady tidal blade load variation through model-scale experiment. It was found that the standard deviations of thrust load range between 200% and 637% of condition without waves. This results in an increase of predicted fatigue damage between 6% and 18%. It was observed that the centre of effort shifts towards the blade root when encountering wave crests of opposing waves, which has not been reported in the literature to date. To reduce errors in fatigue test while the centre of effort is fixed, matching blade shear forces should be sacrificed to match target bending moment at the root. Matching blade shear forces leads to a reduction of predicted fatigue damage ranges from 17% to 25%, which can induce errors in fatigue testing. We anticipate our findings would facilitate the development of fatigue testing of tidal turbine blades.