Biodiversity monitoring surveys are rarely optimised statistically before being initiated. Here, we optimised the monitoring of plants in a temperate forest. The total inventory cost, the number and size of quadrats were optimised to detect a 10% change in species richness over 5 years with alpha = beta = 0.05, using data from ongoing long-term floristic monitoring programs. The procedure showed that the inventory cost would be ca 15% lower using 100-, 200-m2 quadrats instead of 300- or 400-m2 quadrats. Despite the cost associated with the optimisation (e.g. gathering preliminary data) and the imprecise estimates (due to the typically small sample size of the pilot studies), optimisation would often be a better option than expert opinion when designing a monitoring survey.