This paper approaches biological and economic risks in association with strategic and tactical decisions on herbicide dose. Underlying sources of the risks investigated are weed seed bank density and the season-to-season variations in weather. In a simulation model these affect the interplay of dryland wheat crops, a selective post-emergence herbicide and wild oats for different economic and biological outcomes. For fixed and factor-adjusted herbicide dose strategies, our simulation model shows how weather variations result in measurable risks, expressed as cumulative probability distributions of herbicide efficacy, crop yields, changes in weed seed banks in the short and long runs, and economic benefits. Where it is possible to accurately determine the weed densities and weather factors for each spray application, a ‘best efficacy-targeting strategy’ (BETS) is defined. Simulated results with BETS, run with a long sequence of weather, are better than or equal to best fixed doses for any given weed density in terms of crop yields, weed seed bank reduction and long-run economic benefits. Compared to a strategy of continuous maximum doses, BETS allows for lower over-all herbicide use: 23% less with 128 weeds m-2, 58% less with 32 weeds, and 80% less with 8 weeds.