This short article provides a kind of egocentric and longitudinal state of the art of a geographical questioning on the size and evolution of cities. The different stages in the construction of an evolutionary theory of systems of cities are recalled, punctuated by the questions that gave rise to it and by the methods that made possible its formalization. The investigations were at first based on the statistical analysis of empirical data bases about the geographical distribution of urban growth and the accompanying socioeconomic transformations. Stylised facts were identified and used for revisiting the classical models of the hierarchical and functional organization of urban systems and enriching them with dynamic versions. Multi-agent modelling first made it possible to reconstruct the demographic and economic trajectories of cities. Then, artificial intelligence methods (evolutionary algorithms and intensive computing) were used to test the main hypotheses of the evolutionary theory. Although presented here from a personal perspective, all steps of research were produced through fruitful collective work.