The objective of this study is to test the hypothesis suggesting a correlation between the growth rate of gross regional product (GRP) and its initial level (β-convergence). Geographically weighted regression (GWR) was chosen as an analysis tool due to the significant spatial hetero- geneity of parameters. Matrices of distance, trade and migration flows between regions were used as weights. The Akaike criterion (AIC) and the cross-validation criterion (CV) were used to calculate the best number of nearest neighbours (bandwidth). It appears reasonable to group Russian regions in western and eastern clusters. The study shows the existence of unconditional β-convergence and σ-convergence in the western regions. The evaluation of conditional β-convergence model evidences that the convergence is slower in the regions with a larger share of the public sector. The use of trade and migration flows matrix instead of the distance matrix allowed to have more certain results, however, the convergence rate of regions is moderate.