A modification of the GLEAMS model was used to determine application windows which would optimise efficacy and environmental safety for herbicide application to a forest site. Herbicide/soil partition coefticients were determined using soil samples collected from the study site for two herbicides (imazapyr, K,=46, triclopyr ester, K,,=l038)andpublishedvaluesfortwootherherbicides(henarinone,K,=54; triclopyr amine, K,=20) were used in the model. Other site-specific characteristics were taken from catchment topographic maps and soil data. Long-term climatic records for the regionwerethenusedtoprovidemeteorologicaldataforuseinthehydrologycomponent of the model. The model was run with herbicide application for each day of the manufacturers’ recommended growing-season application windows. Average surface run-offlosses,expressedasapercentageofapplied,were low for all herbicidesmadelled (hexazinone, 0.37%; imazapyr, 0.34%; triclopyr amine, 0.21%; hiclopyr ester, I .85%). Model predictions of herbicide loss for each application day were then summarised and the application days with the lowest predicted loss within the manufacturers’ application New Zealand Journal of Forestry Science X(1/2): 28&297 (1996) Michael ef a/.-Using a hydrological model for herbicide application 289 windowswereidentitiedas theenvironmentallysafest days(environmental window)for application, ThisapplicationoftheGLEAMS modelpredictsanenvironmental window, for each herbicide for the site under consideration, during which the probability of adverse environmental impacts is at the lowest level achievable based upon long-term climaticrecords.Theenvironmental window forpesticidesotherthan herbicidescanalso be determinedthrough modelling in a similar fashion. GLEAMS is the model chosen for this example, but other pesticide fate models may be equally applicable.