This paper tends to explore the underlying factors that predict qualification improvement in Russia, which is important to know in light of modernization of Russia aimed to a knowledgebased society model of socio-economic development. The paper is developed on the hypothesis that wage differentiation within and across occupations should play one of the most important roles in up-skilling as in late-industrial society wage gaps within occupations are supposed to be driven by skill diversity within occupations, which appears to be a result of economic modernization. Drawing on contextual and individual level data, I explore a) whether up-skilling in Russia is a propensity of occupations or not; b) to what extent the probability of training in Russia is determined by payoff gaps; and c) how this effect is varied within and across occupations. Application of multilevel modeling to up-skilling revealed great variation in its probability within qualified non-manual workers (managers, professionals, semi-professionals) and small and even more consistent variation within so-called generic labour (clerks, sales workers and manual labour). Wages differences slightly determine probability of up-skilling taken as individual impact, but its occupational variation is barely significant. That is, we can say that probability of training in Russia is an occupational propensity that means that Russia converts to knowledge-based society but still with a large portion of heterogeneity across the workers expected to forming an “occupational face” of such a society. At the same time upskilling in Russia is determined by a number of individual-level factors (including gender, education, age, self-estimation variables and etc.) – and wage impact on probability of training is more likely to be a part of its individual propensity rather than occupational, what shows that occupational skills diversity represented by the differences in workers’ salaries are unlikely to predict up-skilling. Thus, probability of up-skilling in Russia belongs to institutionally biased processes.