A method for performing in situ nutrient enrichment experiments on natural lake phytoplankton communities was developed and evaluated. One set of experiments in which it was employed was designed to detect limiting nutrients and to provide a basis for predicting future experiment results. Individual species responded differently to different treatments, and interactions among the treatment variables were important in shaping the community responses to mixtures of two or three variables. The most consistent features of the productivity results were incorporated into a 'most probable response pattern,' which was partially validated by a second series of experiments. The second experiment series was also used to test the ability of NTA to stimulate phytoplankton productivity. In a third series of experiments sewage effluents were tested in parallel with N and P. Recommendations for the use of in situ enrichment experiments in eutrophication studies are presented. (Modified author abstract)