The relation between the variability of the average annual level fluctuations in the Japan Sea and in the Okhotsk Sea and the indexes, reflecting the dynamics of hydrophysical processes, is examined by means of auto-and inter-correlation analyses; regression forecasting equations are developed as well. Sea level magnitudes are predicted for advance time 1 to 4 years. The sea level forecast model is the multiple linear regression.