Energy　maximising　(EM)　control　of　wave　energy　converters　(WECs)　is　a　noncausal　problem,　where　wave　prediction　information　can　be　used　to　increase　the　energy　conversion　rate　significantly.　However,　current　approaches　do　not　consider　the　prediction　error　evolution　in　the　control　formulation　process,　leading　to　potential　unpredictable　performance　degradation.　Moreover,　most　existing　real-time　WEC　control　approaches　assume　linear　dynamics,　motivated　by　their　simplicity　and　mild　computational　cost　and,　thus,　are　not　effective　for　real-time　control　for　WECs　with　nonlinear　dynamics.　Targeting　imperfect　wave　prediction　and　nonlinear　WEC　dynamics,　this　paper　proposes　a　computationally-efficient　nonlinear　MPC　(NMPC)　scheme　for　WECs　with　(typically)　imperfect　wave　excitation　preview.　This　is　achieved　by　introducing　an　input　move　blocking　scheme　when　formulating　and　solving　the　online　optimisation　problem,　i.e.,　defining　finer　discretisation　grids　for　the　control　input　and　wave　prediction　at　the　early　stages　of　the　prediction　horizon,　where　the　wave　prediction　is　more　accurate,　and　coarser　grids　at　the　latter　stages　of　the　horizon,　to　reflect　less　inaccurate　wave　prediction　information.　Numerical　simulation　results　are　presented,　based　on　a　conceptual　nonlinear　point-absorber　WEC,　to　verify　the　efficacy　of　the　proposed　NMPC　method,　in　terms　of　produced　energy,　computational　complexity,　and　robustness　against　wave　prediction　inaccuracy.