The conditional convergence hypothesis is a major implication resulting from neoclassical growth models. The fact that poorer regions develop faster than richer ones does not have to mean that the variation in their incomes diminishes over time Classical methods of convergence analysis are unable to grasp the income polarisation effect, i.e. "club convergence", either. For this reason, an analysis of full distribution of income and its dynamics may be of interest. This method has been applied to analysis of convergence of regional GDP per capita in Poland in the years 1990-2001. During the analysed period, regional convergence was not the case in Poland, as the GDP per capita was very stable. Nevertheless, one can talk about income polarisation in that period. Relative differences between poor and rich regions were growing. However, the situation in the first half of the 1990s was distinctively different from that in the second half of the decade. In 1990-1995, disproportions diminished sharply. With an overall decline in GDP, richer voivodships were subject to a relatively stronger impoverishment. Nevertheless, one can talk about club convergence rather than absolute convergence during that period. Only during the mentioned sub-period, the poorest regions had the chance to exceed the level of 80% of relative GDP per capita. During that period voivodships faced a much better chance for relative enrichment than impoverishment. Consequently, the club of richer regions was much more numerous than the group of relatively poorer regions. In 1995-2001, the trend was reversed, as the regions were becoming relatively poorer, and income disproportions increased again. The probability for the poorest regions to increase their income above 80% of the average was marginal in that period. The Mazowieckie voivodship, being the richest one, had a clearly distinctive position and was developing much faster than the remaining regions. The tendency towards diminishing the differences in the group of the poorest voivodships, apparent in each of the analysed periods, may be an indication that those regions were becoming relatively poorer and richer alternately as compared to one another, but none of them grew rich enough to break away from the group of the poorest ones.