Bioretention cells (BRC) are used in sewerage systems to retain part of the stormwater runoff and control the flow directed to the system by delaying and reducing the hydrograph peak. In Montevideo with a small rainfall of only 3.6mm some parts of the system overflow. BRC are starting to be used there, applying international guidelines yet without analyses of performance that consider local conditions. The first aim of this study is to develop a BRC model that represents the physical processes of retention and detention. The second aim is to use the local climatic data to assess the performance and test the sensitivity of the results for the input parameters. Then, if possible suggest design guidelines for this location. Two models were developed for a lined and an unlined BRC with Montevideo standard characteristics (rain garden typology), 6 years of recorded rainfall with a 5 min frequency data and design storm events were used. It is concluded that for this BRC typology the parameters of evapotranspiration (ET) and the substrate media characteristics (field capacity, wilting point, and porosity) are insignificant to its performance. The hydraulic loading ratio and the outflow limitation, defined by an infiltration rate or an outlet device, are the key parameters that affect the performance. For Montevideo’s lined rain garden, the retention will only be provided by the ET process, resulting in 6% of the total water budget and no actual detention is accomplished. Therefore, the runoff will enter the sewerage system without any considerable delays and less than 1% of the drainage layer will be used for half of the events, showing this layer is over-designed. For the unlined rain garden, retention was considered as the water that is not directed to the sewerage system (infiltration and ET). For an infiltration rate of 7.2 mm/h (Greenfield criteria) 80% of the total water budget is retained, the remaining 20% overflows for 10% of the recorded events. The mean initial moister content (θi) for the events was calculated using the model and show that for this climatic condition, the available volume for retention before an event is only 24% of its capacity.