Microbial contamination of vegetables due to irrigation with wastewater-polluted streams is a common problem around most cities in developing countries because wastewater is an available source of water and nutrients but wastewater treatment is often inadequate. On-farm treatment of polluted water is a feasible option to manage microbial risks in a multi-barrier approach. Current evidence indicates good suitability of biochar filters for microbe removal from wastewater using the hydraulic loading rate (HLR) designed for sand filters, but their suitability has not been tested under on-farm conditions. This study evaluated the combined effect of several variables on removal of microbial indicators from diluted wastewater by biochar filtration on-farm and the correlations between removal efficiency and HLR. Columns of biochar with three different effective particle diameters (d) were fed with diluted wastewater at 1x, 6x, and 12x the design HLR and two levels of water salinity (electrical conductivity, EC). Influent and effluent samples were collected from the columns and analyzed for bacteriophages (ɸX174 and MS2), Escherichia coli, Enterococcus spp., and Saccharomyces cerevisiae. Microbe removal decreased with increasing HLR, from 2 to 4 to 1 log for bacteria and from 2 to 0.8 log for viruses, while S. cerevisiae removal was unaffected. Effective particle diameter (d) was the main variable explaining microbe removal at 6x and 12x, while EC had no effect. Correlation analysis showed removal of 2 log bacteria and 1 log virus at 3x HLR. Thus biochar filters on-farm would not remove significant amounts of bacteria and viruses. However, the design HLR was found to be conservative. These results, and some technical and management considerations identified, can assist in the development of a scientific method for designing biochar filters for on-farm and conventional wastewater treatment.