Mechanistic studies of cellular processes are usually carried out with large populations of cells. However, parameters that are measured as averages of large populations can be misleading. For instance, an apparently linear response to a signal could, in fact, reflect an increasing number of cells in the population that have switched from 'off' to 'on', rather than a graded increase in response by all the cells. At present, the study of single cells is challenging, but new technologies mean it might soon be a reality.