Herbicide labels recommend sufficiently large doses to achieve high efficacy of weed control under a range of environmental conditions. While users may not apply doses greater than recommended they may apply less, but do so without quantitative guidance. This paper explores long-run biological and economic outcomes of fixed label doses and strategies keyed to density of weeds in Òbest fixed-doseÓ (BFD) and Òbest efficacy-targetingÓ (BET) modes. Analysis based on 59 experiments in wheat fields across southern Australia, from 1986 to 1995, indicate the latter strategies are superior to maximum label rates in terms of (a) mean net present values of current and future benefits and costs of weed management, expressed as Hamiltonians keyed to weed seed banks, (b) mean current gross margins, and (c) lower overall herbicide use. BFD have further practical advantages in simplicity and lower information requirements for the user, BET in tailoring applications to specific field environments.