Deterministic watershed models are often used to estimate the probable maximum flood (PMF). An approach for investigating the uncertainty in extreme flood modeling is proposed. Using different calibration objectives, several automatic calibrations of the University of British Columbia watershed model (UBCWM) are conducted and the resulting collection of optimal combinations of parameter values are used to simulate the extreme event. An application to the Coquitlam River Watershed above Coquitlam Dam in southwestern British Columbia shows that the variability among the PMF estimates is relatively small in comparison with the potential uncertainties in estimating extreme events, with coefficient of variation values for peak flow, event volume, and time to peak of 4%, 1%, and 1%, respectively. The PMF-based simulations are relatively insensitive to the different measures of calibration performances.