The scope of this study was to assess the usefulness of top probability distributions to describe maximum rainfall data in the Lusatian Neisse River basin, based on eight IMWM-NRI meteorological stations. The research material was composed of 50-year precipitation series of daily totals from 1961 to 2010. Measurement data series were supplemented using weighted average method. Homogeneity for refilled data were investigated by precipitation double aggregation curve. Correlation between the measurement data varied from 96 to 99% and did not indicate a disorder in the homogeneity of rainfall data series. Variability of recorded daily precipitation maxima were studied by linear regression and non-parametric Mann-Kendalls test. Long-term period changes at maximum rainfalls for four station remained as statistically insignificant, and for other four were significant, although the structure of maximums were relatively similar. To describe the measured data, there were used the Fréchet, Gamma, Generalized Exponential Distribution (GED), Gumbel, Log-normal and Weibull distributions. Particular distribution parameters were estimated using the maximum likelihood method. The conformity of the analyzed theoretical distributions with measured data was inspected using the Schwarz Bayesian information criterion (BIC) and also by the relative residual mean square error (RRMSE). Among others, the Gamma, GED, and Weibull distributions fulfilled the compliance criterion for each meteorological station respectively. The BIC criterion indicated GED as the best; however differences were minor between GED on the one hand and the Gamma and Weibull distributions on the other. After the conduction of the RRMSE analysis it was found that, in comparison to the other distributions, GED best describes the measured maximum rainfall data.