I use Russian panel data for 1994-2013 to estimate the earnings function that includes age, period and cohort. To eliminate the linear dependence between these variables, I substitute period for the real wage index in the earnings equation. The earnings function is estimated by the Heckman selection estimator separately for males and females. The estimation reveals strong between-cohort differences in human capital, especially for Russian men. The three sources of the between-cohort differences in human capital are detected. The major one is due to the human capital devaluation resulted from the transition of Russia to the market economy in the 1990’s. This source is responsible for the 50% difference between the incomes of Russian men born in 1940-1949 and the salary of men born in 1989. Other two sources of between-cohort differences in earnings are, first, distinct conditions at the moment when workers entered the labor market, and, second, between-cohort differences in return to education.