Using both regression analysis and an unsupervised graphical model approach (never applied before to this issue), we confirm the rejection of the Gibrat's law when our firm-level data are considered over the entire investigated period, while the opposite is true when we allow for market selection. Indeed, the growth behavior of the re-shaped (smaller) population of the survived most efficient firms is in line with the Law of Proportionate Effect; this evidence reconciles early and current literature testing Gibrat's law and may have interesting implications in terms of both applied and theoretical research.