Registration of agricultural chemicals requires an assessment of the potential ecological consequences of the purposeful release of such chemicals into the environment. If preliminary assessments, such as laboratory toxicity tests, indicate that potential unacceptable adverse effects may occur, then aquatic field tests are required. Classically these tests have entailed the treatment of agricultural fields adjacent to aquatic habitats and the subsequent measurement of any effects on the aquatic system. However, problems with factors such as variability led to the development of the mesocosm concept. Mesocosms are artificially constructed ponds, which, because they can be replicated, can provide statistically supportable data on the effects of a chemical. Because mesocosms have not been used for pesticide registration, there are many unknowns about end points, experimental design, dosing, and data interpretation. This manuscript focuses on experimental design and its importance in the methods used for risk assessment. Experimental designs that center around hypothesis testing are compared with regression designs. Although statistically the designs may be similar, the use of the data in the final risk assessment is influenced by the choice of design. In addition, the importance of choosing the most accurate dosing levels will be determined by how the risk assessment is done.