The recommended dose for many pesticides is expressed as a constant mass or volume per unit ground area covered by the crop. This method of dose expression is well suited to boom spraying where a reasonably uniform horizontal distribution of deposit can be achieved with a well-adjusted sprayer. However, in many practical situations (e.g. broadcast spraying of apple trees or other row structures where the spray application is made from within the canopy) the horizontal deposit distribution is strongly influenced by the crop area density and other crop structural parameters. This paper describes a generic method of pesticide dose expression to investigate these effects. The method incorporates a model of the spray volume deposition process. The model assumes that the pesticide deposit is proportional to the tank-mix concentration of pesticide. The model also assumes that spray volume deposit is proportional to the applied spray volume per unit row length and is inversely proportional to a crop length scaling function L (i.e. a parameter with the units of length that is expressed as a generic function of different crop parameters). The useful working range of this model is bounded by the condition for high spray volume where target losses become significant due to saturation and the condition for very low volume where evaporative transport losses become significant. Within this framework, four different models are formulated using first-order approximations for the length-scale as functions of the following crop parameters: tree row spacing, tree row height, tree area density and tree row volume to ground area ratio. Published measurements of crop structure and spray volume deposit on apple trees are compared with the output from these models. Light detection and range (LIDAR) measurements of apple orchards are presented and used in conjunction with the different models to predict pesticide use associated with different methods of dose expression. The results demonstrate the relative potential for varying the pesticide application rate according to the different crop parameters. The results enable the identification of reference orchards that could be used to establish worst-case pesticide application rates for registration purposes. The results also enable the identification of other orchards and growth stages where pesticide application rate might be reduced by up to a factor of five and give the same pesticide deposit as the reference structure.