Mathematical modeling forms a large part of modern ecological research. This essays argues that it forms too large a part. The uses of modeling are reviewed: models are used to "explain" the functioning of natural ecosystems; they are also used in forecasting, in hypothesis-generation, and as standards of comparison. Although the original task of models was to explain nature, this is the task they perform least well. To compare a model with reality is, in effect, to ask a number of questions simultaneously when they could more profitably be tackled be tackled individually. The search for answers to single "unit" questions, here called "investigating," is more likely than modeling to yield advances in knowledge. It is not easy, as examples show.