In the experiments reviewed in this article the subjects are asked to produce ideas that are relevant to a given task request (e.g., possible consequences of a hypothetical event). After describing the specific task material and the performance measures used in the relevant research studies, some analytic background is given by outlining the cognitive resources required in this kind of experimental task and by listing the various factors that may come into play when subjects perform in groups (with discussion) instead of individually. We then review the studies comparing individual and group performance. In all of these experiments the subjects were asked to work according to the rules of brainstorming, which prescribe that participants refrain from evaluating their ideas. This procedure purportedly results in superior group, relative to individual, performance. However, the empirical evidence clearly indicates that subjects brainstorming in small groups produce fewer ideas than the same number of subjects brainstorming individually. Less clear evidence is available on measures of quality, uniqueness and variety. The discussion considers factors that may be responsible for this inferiority of groups. The role of social inhibition receives particular attention also in terms of suggestions for research. Apart from the group-individual comparison we review the existing research concerning factors that may influence group performance on idea-generation tasks.