A critical review of the mechanisms of present water treatment systems including, chlorination, coagulation, filtration, granular and powdered activated carbon adsorption, ozonation and ultraviolet radiation for the removal of herbicides is presented. Rapid selective and sensitive HPLC methods were developed and rigorously validated for the analysis of the selected herbicides. Analysis of atrazine was made using Cl8 cartridges. For raw water containing interferences, extraction of the compound was made on SCX cartridges, followed by solvent exchange on C18 cartridges. A quantitative recovery of virtually 100% of the compound was achieved using C18. While the double cartridge extraction of the compound gave a recovery of about 89%. Previously developed methods for 2,4-D and MCPA were rigorously validated for the extraction and analysis of 2,4-D and MCPA. A quantitative recovery of usually greater than 90% was achieved for both compounds using Cl8 cartridges. For the extraction of paraquat different extraction systems including, reversed phase on C8 and C18, ion-paired reversed phase on C18, and cation exchange on SCX, CBA, and CN were investigated. A quantitative recovery, usually greater than 90%, of the compound was obtained using CN and CBA cartridges. The methods were then successfully used for the evaluation of the removal efficiency and establishment of mechanisms of removal of herbicides by biological filters at bench and pilot scale. Four herbicides belonging to three broad chemical categories were studied. The data presented in this study demonstrated that biological filters are very efficient in removing certain classes of herbicides. 2,4-D and MCPA were consistently removed to below a detection limit of 0.1 mug/1 for an influent concentration of 3-11 mug/1. Process variables such as flow rate, bed depth and contact time were investigated for the efficient removal of these herbicides. Seasonal variations in performance were observed and possible explanations proposed. A series of experiments was undertaken to establish .mechanisms of removal. Quantitative recovery of the herbicides from the river water proved that the processes in the filter bed as opposed to the processes in the water were responsible for removal of the herbicides. It was clear from the investigation of the adsorption of the herbicides both on the sand and organic and inorganic dirt that adsorption on these surfaces was not the main reason for removal. Filter maturation experiment showed that the presence of microorganisms in the bed is a precondition for the removal of herbicides. A depth experiment for the removal of 2,4-D showed that superficial efficient zone of removal imitates the distribution of microbial density. This evidence confirms the significance of microorganisms for the removal of herbicides by the filter bed. The ultimate proof of the biodegradation of 2,4-D by microorganisms in the filter bed was the identification of the biodegradation product 2-chlorophenol as predicted by the metabolic pathways of the compound. Filter design modifications using activated carbon were made to accommodate the removal of 'non-biodegradable' herbicides. A sandwich sand / GAC / sand filter was investigated. Filter efficiency for this arrangement was determined and short-comings were identified and a possible solution in the form of a double GAC sandwich is suggested.