We applied Bayesian networks to distinguish groups of countries different in terms of their level of entrepreneurship growth as measured by new business density and to find factors responsible for the differences. The objective of the article is to check the applicability of Bayesian networks in filtering variables critical for entrepreneurship and its growth in the multidimensional data environment, and hence to shed a new light on the forces underlying the phenomenon. We detected considerable differences in the sets of key factors responsible for entrepreneurship growth among five groups of countries sorted out on the basis of their performance in terms of the entrepreneurship growth. Except the positive dimension, our study has also a normative value as the empirical findings gathered by us can help to formulate fitted recommendations as to the entrepreneurship stimulation in various socioeconomic and political settings.