Water samples were collected from a constructed wetland in Armstrong Creek, Geelong, for eight storm events. For each sampling, a single surface sample was collected at the deepest point of five equidistant transects across the wetland in the 12, 36 and 60 h following the storm events. The samples were tested for total suspended solids (TSS), total nitrogen (TN) and total phosphorous (TP). The results from the five transects were averaged and used to develop an empirical exponential decay model for the wetland. The wetland model was integrated within a catchment model based on the 1-D Stormwater Management Model (PCSWMM), which estimates the runoff from the contributing catchment into the wetland. The catchment model was calibrated/validated using water quality data collected from 8 storm events and flow data measured over a period of 22 months at sites located in Geelong. The catchment model was validated against water level measurements in the wetland showing good agreement with a correlation coefficient of 0.78 (p < 0.05). The wetland water quality model was validated against water quality data obtained from field sampling with correlation coefficient of 0.749 (p < 0.05), 0.491 (p < 0.10) and 0.856 (p < 0.05) for TSS, TN and TP respectively. A full year simulation was run using rainfall obtained for 2018 (assumed to be a typical year) to assess wetland performance. The model results show that annual loads (kg/yr) retained were 66%, 55% and 50% for TSS, TN and TP, respectively. These results exceed the design retentions of 45% for TP and TN, but not the 80% required for TSS as stipulated by the Victorian Stormwater Quality Objectives (2017). Further modelling, increasing the detention volume by raising the overflow weir crest elevation by 0.5 m resulted in an additional 6% retention in TSS. This is still insufficient to meet the 80% target. The main conclusion of this study is that through the proposed combined sampling and modelling approach, the annual wetland pollutant treatment levels within the constructed can be accurately identified. As water quality assets are designed based on annual treatment levels, this approach provides a cost-effective method of appraising wetland performance.