Global flood losses are projected to continue rising as a result of climate change and development patterns. Standard methods for regulating development along floodplain corridors rely on deterministic flood hazard boundaries that assume stationarity and neglect inherent uncertainties associated with approximating complex processes. Consequently, there is a need to quantify the uncertainty in flood hazard estimates to avoid inadvertently placing an increased portion of the population at an elevated risk. The goal of this research is to develop improved methods for quantifying and portraying flood hazards that communicate the uncertainty in model estimates, account for nonstationary climate and land use, provide meaningful descriptions of flood hazards, and serve as a template for floodplain management and land use planning. I developed the first probabilistic floodplain mapping framework that simulates uncertainty in streamflow, land use, and geomorphic adjustment through Monte-Carlo simulations (MCS) of flood hydraulics while accounting for nonstationary flood peak distributions. Flood inundation likelihood resulting from MCS revealed spatial heterogeneity in probabilistic gradients and locations of elevated or hidden risk that were not apparent with deterministic approaches. A novel simplified uncertainty bounding (SUB) method that provides a parsimonious alternative to conducting MCS was also developed and evaluated to increase the practicality of uncertainty analysis for a larger user base. The SUB method generally provides reasonable approximation of uncertainty derived from MCS but tends to overestimate uncertainty and thereby produce a precautionary estimate of flood risk. The performance and acceptability of its performance will depend on the application, as hydraulic structures and valley shape can have a significant effect on the accuracy of non-MCS uncertainty estimates. Reliability over a planning horizon has been suggested as an improved metric for quantifying and communicating flood risk, and it is typically reported as an expected likelihood that realized flood magnitudes will not be exceed a design value. However, reliability is a random variable dependent on uncertainty in the joint probability distribution of annual flood peaksand channel capacity. I built upon existing research to provide the first evaluation of the compounding effects of multiple sources of uncertainty and nonstationarity on the reliability of flood protection measures. This research provides the first probabilistic floodplain maps to illustrate the spatial variability and uncertainty in reliability distributions derived from local channel and floodplain characteristics, and it points to the feasibility of operationalizing these approaches in practical applications. The results of this research illustrate the need for a paradigm shift in floodplain management to a more transparent depiction and communication of flood risk under uncertainty.