Two outdoor subsurface flow beds (control and experimental, 10 m x 1 m) were filled with a substrate of pea gravel (3-6 mm) to a depth of 60 cm. The experimental bed or small-scale constructed wetland was originally planted with Typha seedlings at a density of 7.5 plants/m2. Both beds (experimental and control) were treated with the same aqueous concentrations of diesel oil under identical dosing conditions. The average overall hydrocarbon removal efficiencies at the three monitored depths (top, middle and bottom) in the subsurface systems were 80.1 +/- 9.8%, 78.0 +/- 9.1% and 71.6 +/- 10.0% in the experimental bed and 72.3 +/- 11.9%, 69.1 +/- 10.3% and 63.4 +/- 9.4% in the control bed. The differences in the hydrocarbon removal efficiencies between corresponding months in 1999 and 2000 were statistically analysed and are generally not significant. The individual hydrocarbon removal efficiencies exceeded 60% in the top sections of both beds except for C-11 and C-25 with C-23 and C-26 also reduced in the control bed. Overall differences in the removal efficiencies of the planted and the unplanted beds as well as at different depths in both systems, indicate that Typha related removal processes complementing adsorption onto the gravel substrate are occurring.