Ten trials were established in the 1986 vintage year on vineyards in three vine‐growing districts of New Zealand, to examine changes in the frequency of resistance to dicarboximide fungicides of the Botrytis cinerea population under different fungicide programmes. The vineyards chosen encompassed sites with pre‐season resistance frequencies ranging from 0 to 97%. Under a non‐dicarboximide programme, using dichlofluanid and chlorothalonil, resistance frequency had decreased at harvest irrespective of the initial frequency. In programmes incorporating dicarboximides (chlozolinate, iprodione, procymidone, vinclozolin), the resistance frequency tended to increase if pre‐season resistance frequency was low, and decrease if it was high. The hypothesis is presented that for a given dicarboximide programme, resistance frequency tended to one value, the balance value, where selection for resistant strains due to fungicide application balanced selection against the resistant strains due to their low fitness. This value was estimated for the 1986 vintage year to be 38% for a programme where application of dicarboximides was restricted to the period from the veraison (sugar rise), and 64% where dicarboximides were applied from the pre‐bunch closure phase. Actual values observed at harvest 1987, after these same programmes had been applied for two consecutive years, were 39% (S.E. 8‐5%) and 68% (S.E. 8‐8%), respectively.