The rank-size model - which states that the size distribution of cities in a country follows a Pareto distribution - has been recognized as one of those stylized facts or amazing empirical regularities, in spatial economics. A common problem in city size distribution studies concerns the definition of “cities”, namely the consistency of those definitions over time. In this paper we use a city-proper data base which uses a consistent definition of cities from 1864 to 1991. Portugal is a country with long established national borders and whose mainland urban system shows a constant number of cities over that period. In Portugal, empirical evidence on city size distribution based on census data shows that two large cities dominate the urban system, associated with a large number of very small cities and a clear deficit of medium-size cities. In this paper we analyse the evolution of the rank size exponent and examine the effect of varying city size cut-offs on the estimate value of that exponent. Then, we study the deviations of the rank-size distribution from linearity. Finally, we explore the dynamics underlying the evolution of the urban system by examining the relationship between city growth rates and city size.