In the context of climate change and urbanization, sustainable urban drainage systems (SUDS) are widely adopted measures to manage stormwater in the city on-site. However, their performance in practice often differs from modelled and laboratory-scale predictions due to the variability in properties of real sediments (in terms of size, shape, density and coagulation) compared to the silicate standard Millisil®W4. Clogging is a common source of failure.  The SediSubstrator L is a decentralized stormwater treatment device installed as a pre-treatment step to mitigate clogging in a storage and infiltration system on the Rooseveltlaan in Amsterdam. It consists of a sedimentation pipe with a flow-separating grate, the SediPipe, and a filter-adsorbent, the SediSorp+. It is purported to remove 80 % of TSS by DiBT (the technical authority in the German construction sector) test principles that use Millisil®W4 to simulate real sediments. The full-scale unit was monitored in the city throughout May-September 2022 to assess its performance.  The stormwater runoff discharged from the catchment had high concentrations of lead (54 μg/L) and zinc (790 μg/L), likely due to contact with gutters and old roofing material, amplified by the relative contribution of these roofs to the total catchment discharge (accounting for 50 % of the area contributing to runoff). The sediment (TSS) concentration was low, equivalent to 20 mg/L on average. The sediments were also light and fine—with an organic fraction of 66 % and with 78 % of diameter smaller than 63 μm. In the SediSubstrator L, the TSS removal efficiency was 34 % on average. This corresponds to an estimated caught load of 2.7 kg for this period. The removal efficiency was shown to increase with an increasing stormwater TSS concentration, with longer antecedent dry periods and with lower TSS organic fractions. Turbidity dynamics in the system suggest that while a net sequestration of solids occurs in the SediPipe, there is a resuspension of fine solids. This was observed in a camera inspection to occur from solids which settle on or near the grate. In an extreme rainfall event on September 28th 2022, water collected on the section of the street connected to the SediSubstrator, the cause of which is still subject to speculation. The observed SediSorp+ filter resistance across the summer was not indicative of gradual clogging, but an inspection showed signs of decayed organic matter throughout the full length of the filter bedas well as traces of cement in two of the four cartridges. It is possible that these two effects together with turbulent inflows prompted the acute clogging behavior. There is interest in using the SediSubstrator beyond the city of Amsterdam to reduce phosphorus loadings in the road runoff discharged to sensitive nature areas. On the Rooseveltlaan, the average total phosphorus removal efficiency was 18 % (50 % for dissolved, readily bioavailable ortho-phosphate). Interactions with settled sediments generated ortho-phosphate in the SediPipe, and fine particulate and colloidal organic phosphorus was shown remobilized in both the SediPipe and SediSorp+. The removal of ortho-phosphate in theSediSorp+ in natural rainfall was good (on average 50 %) and was shown to be consistent at different contact times (approximately 10-30 minutes).  The installed unit should be monitored over a longer time period of two years for statistical significance and to capture seasonal variation in loads. Nevertheless, the removal efficiency observed on site is consistent with the results of a sedimentation model developed according to Ferguson & Church (2004), using a stormwater sediment particle density as measured at another location in the city. Design adaptations are recommended to improve the SediSubstrator L to the conditions observed in Amsterdam: namely, better site selection, a longer SediPipe section (24 m) and a second filter stage to better capture the fine suspended solids.