Abstract Procedures used to detect environmental impacts that occur as a result of planned disturbances are often inadequate. Widely used designs for univariate measures, such as the abundance of a population, lack proper spatial replication and have unjustified patterns of temporal sampling. Asymmetrical analyses of variance derived from repeated measures models can be used to detect many types of impact that are not identifiable using widely recommended BACI (Before/After, Control/Impact) sampling. These asymmetrical, beyond BACI designs are also more logical because of spatial replication. The mechanics of these procedures are discussed, including worked examples of calculations, considerations of their power to detect impacts of a specified magnitude and the integration of various temporal and spatial scales into the design. Related issues are briefly discussed concerning optimization of sampling and how to proceed when no data are available before a disturbance.