This essay is writtened to clarify the difference of saving pattern between wage-earners and non-wage earners. In our country, the family budget data in 28 cities has been shown monthly. But as for non-wage earners, the income data has not been researched. If the summation of expenditure to consume, saving and tax is equal to income, we can easily compare the saving pattern of wage and non wage earners. But the family budget data do not deal with net saving, - disposable income minus consumption -, but the gross saving that is paid in that month. So that if we estimate the non-wage earner's income in above descriptive method, we shall meet some statistical error. We tested the degree of this error by wage-earner's family budget data, and compare the saving pattern of wage and non-wage earners. So far as family budget data is concerned, the propensity to save of wage earners is larger than non-wage earners at 1956. On the other hand, according to predictional research to consume, the wage-earner's propensity to save is smaller than non-wage earners. What is the cause of this difference? First the fitness of the regression line is not always good. Although the co-efficient of correlation is relatively high, but in some cases we ought to regard the correlation as curve, not a line. Saving is affected not only by income level, but also income change, liquid assets, labor force etc., Second, according to predictional research, the "saving" is dealt with as net saving. In national accounting, we deal with the aggregate saving. But owing to the deficit of higher income group's family budget data, we cannot estimate the aggregate saving pattern. In place of that analysis, we consider the relation between the distribution of banking deposit and income, and conclude; (1) we may guess that the propensity to save higher income groups will be increasing progressively. (2) the relation between saving and liquid assets is not simultaneous. The larger the amount of liquid assets are, the larger the saving-ratio is.