Queensland state in Australia often suffers from devastating floods. To design flood-safe infrastructure at ungauged catchments, regional flood frequency analysis (RFFA) method is adopted, which transfers flood characteristics from gauged catchments to ungagged site. This study evaluates quantile regression technique (QRT) for 10-year (Q10) and 50-year (Q50) return periods. Diagnostic plots confirmed that both models satisfied the assumptions of ordinary least squares, with residuals randomly distributed and over 90% within ±2, indicating stable variance and approximate normality. The Q10 model demonstrated stronger predictive skill, with an R2 of 0.71 compared to 0.61 for Q50. Error analysis showed lower mean and median absolute relative errors for Q10 (39.41% and 26.00%) than for Q50 (54.93% and 30.92%), along with reduced variability, highlighting greater reliability at shorter return periods. Further study should include a higher number of predictor variables, larger data set and non-linear methods to develop and test RFFA models.