This is an in-depth uncertainty analysis of Artificial Neural Network (ANN)-based Regional Flood Frequency Analysis (RFFA) in the Australian context. This uses data from 88 gauged stations in New South Wales, Australia. Using eight hydrological and physiographical predictors, six design flood quantiles (Q2, Q5, Q10, Q20, Q50 and Q100) were modelled. Model robustness was tested through a Monte Carlo simulation technique with random 70/30 train–test splits, producing a distribution of model outcomes across simulation runs. Uncertainty was quantified using the median relative error ratio (REr), yielding values of 59.62 ± 10.73% (Q2), 55.86 ± 10.6% (Q5), 54.13 ± 10.76% (Q10), 55.86 ± 11.44% (Q20), 57.72 ± 12.37% (Q50), and 59.81 ± 10.58% (Q100). Findings show that ANN performance varies with return period, with mid-range quantiles generally achieving lower uncertainty. These results demonstrate both the potential and the limitations of ANN-based RFFA, reinforcing the need for uncertainty analysis before applying AI-driven RFFA models to ungauged catchments.