Constructed treatment wetlands can be an effective and sustainable method to remove pathogens that pose health risks from agricultural runoff. This study evaluated the removal of fecal coliform (FC) from agricultural runoff in a lightly loaded surface-flow treatment wetland prior to discharge to the Yakima River, Washington State, USA. The 1.6 ha system consisted of a sedimentation basin (1.4 d hydraulic retention time) followed by two wetlands (5–6 d hydraulic retention time). FC in inflow ranged from 100 to 1,000 cfu/100 mL. Mean annual FC log-removal in the sedimentation basin was 0.66 ± 0.17 (mean plus/minus standard deviation; n = 7). However, there was a comparable production of FC within the two wetlands where annual log-removal averaged −0.71 ± 0.39 in the north wetland and −0.57 ± 0.17 in the south wetland. FC removal in the sedimentation basin weakly correlated with turbidity removal (R2 = 0.13, p < 0.01, n = 61), suggesting that settling was an important FC loss mechanism. FC removal in the wetlands negatively correlated with temperature (R2 = 0.27–0.33, p < 0.01, n = 26) indicating that survival and/or reproduction was an important FC production mechanism. Muskrat colonization in the wetlands in 2007 and 2008 corresponded with a marked increase in FC in wetland outflow. Results suggest that, regardless of the presence of muskrats, sedimentation basins alone are more effective than a combined sedimentation basin–wetland system in removing FC from dilute agricultural runoff.