Predicting the consequences of extinction is an important goal in ecology, due to the accelerating rate of global and local species loss. Cascading extinction~ the extinction of one species as a consequence of the extinction of another, has been demonstrated in several natural ecosystems and is widely predicted by theoretical studies. Extinction may also affect the functioning of ecosystems, as functionally important species may be lost from communities and positive interactions between species may be disrupted. Investigating the consequences of extinction in natural ecosystems is difficult, as removing a species from an area of habitat is often problematic. Also, because indirect effects may take many generations to be revealed, experimental manipulations in natural ecosystems often need to last for several years. Experimental aquatic microcosms provide a new opportunity for conducting removal experiments in small, tractable, communities containing rapidly reproducing species. Here, I present the results of four experiments conducted in such microcosms that . investigated the consequences of the removal ofa species and the effect of temperature and energy on patterns ofextinction. In general, local diversity was robust to the removal of a species. No cascading extinction was observed and some positive effects of extinction for diversity were revealed. Species loss had a positive, negative or no effect on ecosystem functioning, depending upon community composition. Small changes to the identity of the species present in communities had large effects; the consequences of extinction for . . ecosystem functioning was very context-dependent and rare species were important. Temperature and chemical energy interacted to determine species diversity and extinction determinism, making it difficult to predict the consequences of changes to either temperature or chemical energy for extinction without knowledge of the other.