SummaryThe dose responses of 14 wild plant species (two grasses, two legumes, one annual and nine perennial dicotyledons), not usually recognised as weeds, to four herbicides (asulam, glyphosate, MCPA and mecoprop) were measured in glasshouse experiments. Glyphosate was the most toxic; seven of the species tested had EDu, values (measured as shoot dry weight) of<1.0 μg/plant, compared with only one species for MCPA and mecoprop. Asulam was the least toxic.Results were used to indicate the risk to each species from drift damage. A model of spray drift, based on that developed by Thompson&Ley (1982) for evaporating droplets, was rescaled to allow for field application rates and used to predict the distances travelled by given doses of herbicide. This gave acceptable agreement with reports for drift damage in the field, and predicted that only glyphosate sprayed at the highest recommended concentration might be unsafe to some of the species examined. The two herbicides sometimes used as volatile formulations (MCPA and mecoprop) did not cause damage at the small doses likely to result from exposure to vapour in the field