Two free water surface (FWS) and two subsurface flow (SSF) pilot-size constructed wetlands treating highway runoff (HRO) were monitored over a period of two years (September 2005-August 2007). One FWS and one SSF were designed with a hydraulic retention time (HRT) of 12h, named FWS12 and SSF12, respectively, with each one capable of treating a maximum HRO of 12.6 m(3) d(-1). The other couple, named FWS24 and SSF24, respectively, was designed with an HRT of 24h, with each receiving a maximum HRO of 6.3 m(3) d(-1). The influent flowed from a highway section with a total surface 2752 m(2) on the island of Crete, Greece, in the heart of the South-Central Mediterranean region. Influent and effluent were monitored for COD, TSS, total N (TN), NO(3)(-) and total P (TP) concentrations. Furthermore, removal efficiencies were examined for heavy metals (Cu, Ni, Pb, Zn) for both years, while polycyclic aromatic compounds (PAHs) were examined for the period between September 2006 and August 2007. The influent had a two-year average COD value of 101 mg l(-1), whereas the mean values for TSS, TN, N-NO(3)(-) and TP were 203, 4.30, 1.25 and 4.17 mg l(-1), respectively. For Cu, Ni, Pb and Zn the respective two-year mean influent concentrations were 56, 114, 49 and 250 microg l(-1). Mean concentration of total PAHs in runoff (summation operator PAHs, 16 compounds) were 12.01 microg l(-1). The performance among the four beds was not significantly different according to ANOVA analysis followed by Tukey test (at p<0.05) for almost all the above physicochemical parameters, suggesting that all systems performed in a similar way. All studied systems, achieved a mean of two-year removal efficiencies of 47% for COD, 89% for TSS, 49% for TN, 58% for N-NO(3)(-), 60% for TP, 47% for Cu, 23% for Ni, 33% for Pb, 61% for Zn and 59% for summation operator PAHs (16 compounds).