The author has analysed in the previous paper the fact that statistical distributions of heights of waves are very complicated and various owing to the sea states and weather conditions, and that by a parameter n, where n>2 represents “regular swells”, n<2 “violent seas”; and for the case of n=2 is “confused seas and swells” which had already pointed out by Drs Longuet-Higgins. Statistical amplitude of rolling of a ship on such sea surfaces are : θ=F (βs) ·Φ ( n ) ×θ0, where θ0 in perfect resonance amplitude due to regular waves, F (βs) and Φ ( n ) is factor of correction due to irregularity of waves, namely [F (βs)] n=βs∫ xs 0 [ I (x) / I (O) ] n dx , I (x) ≅x2e-βsx2/√ (x2-1) 2+2Nsγsδmx2e-βsx2, Φ ( n )=q [Γ (1+1/n) -n√lnq∞j=1∑ (-1) j (lnq) j/j! (jn+1)] + (1-q) n√lnq (average value) Γ (1+1/n) [1-Nj=0∑ (-1) jNCj/n√j] or n√lnN [1+1/n∞j=1 (-1) jj-1S=0Π (1-sn) ·1/j (lnN) j] ∫∞0e-z (lnz) jdz+etc, (maximum value) q=3 corresponds to Drs. Sverdrup-Munk's significant mean value, N is numbers of swings of rolling. Ns is damping coefficient, γs is effective slope coefficient, βs wave age for reasonance ; and xs= Tw/Ts, Tw is maximum or most frequently observed periods of waves, Ts is ship's natural period of rolling. δm=0.140 is the maximum value of wave steepness. Here the author assumes that density of wave spectrum between β and β+ d β are 1/2ρg r 2 (β), r (β) =π/g (βu) 2δ. and spectral steepness δ (β) is the same that of wave steepness. In the estimation I (x) the author used Dr. Neumann's steepness curve δ=0. 140 e-β2. Cumulative wave energy is then expressed by the following formula E=∫βm01/2ρg r 2 (β) d β, βm=gTw/2πu where βm can be analysed by Drs, Sverdrup-Munk's theory. u is wind velocity