In order to evaluate the influence of substrate composition on stormwater treatment and hydraulic effectiveness, mesocosm-scale (180 L, 0.17 m(2)) laboratory rain gardens were established. Saturated (constant head) hydraulic conductivity was determined before and after contaminant (Cu, Zn, Pb and nutrients) removal experiments on three rain garden systems with various proportions of organic topsoil. The system with only topsoil had the lowest saturated hydraulic conductivity (160-164 mm/h) and poorest metal removal efficiency (Cu ≤ 69.0% and Zn ≤ 71.4%). Systems with sand and a sand-topsoil mix demonstrated good metal removal (Cu up to 83.3%, Zn up to 94.5%, Pb up to 97.3%) with adequate hydraulic conductivity (sand: 800-805 mm/h, sand-topsoil: 290-302 mm/h). Total metal amounts in the effluent were <50% of influent amounts for all experiments, with the exception of Cu removal in the topsoil-only system, which was negligible due to high dissolved fraction. Metal removal was greater when effluent pH was elevated (up to 7.38) provided by the calcareous sand in two of the systems, whereas the topsoil-only system lacked an alkaline source. Organic topsoil, a typical component in rain garden systems, influenced pH, resulting in poorer treatment due to higher dissolved metal fractions.