Estuarine compound flooding can happen when an extreme sea level and river discharge occur concurrently, or in close succession, inundating low-lying coastal regions. Such events are hard to predict and amplify the hazard. Recent UK storms, including Storm Desmond (2015) and Ciara (2020), have highlighted the vulnerability of mountainous Atlantic-facing catchments to the impacts of compound flooding including risk to life and short- and long-term socioeconomic damages. To improve prediction and early-warning of compound flooding, combined sea and river thresholds need to be established. In this study, observational data and numerical modelling were used to reconstruct the historic flood record of an estuary particularly vulnerable to compound flooding (Conwy, N-Wales). The record was used to develop a method for identifying combined sea level and river discharge thresholds for flooding using idealised simulations and joint-probability analyses. The results show how flooding extent responds to increasing total water level and river discharge, with notable amplification due to the compounding drivers in some circumstances, and sensitivity (~7 %) due to the time-lag between the drivers. The influence of storm surge magnitude (as a component of total water level) on flooding extent was only important for scenarios with minor flooding. There was variability as to when and where compound flooding occurred; most likely under moderate sea and river conditions (e.g. 60–70th and 30–50th percentiles), and only in the mid-estuary zone. For such cases, joint probability analysis is important for establishing compound flood risk behaviour. Elsewhere in the estuary, either sea state or river flow dominated the hazard, and single value probability analysis is sufficient. These methods can be applied to estuaries worldwide to identify site-specific thresholds for flooding to support emergency response and long-term coastal management plans.