People ordinarily distinguish between ways of behaving that are "normal" and those that are "abnormal." But how exactly is this distinction to be understood? This talk will discuss a series of experimental studies designed to explore people's ordinary notion of normality. The key result is that people's ordinary notion of normality is not a purely statistical one (e.g., the type of behavior that is most frequent) or a purely prescriptive one (e.g., the type of behavior that is ideal). Instead, the ordinary notion of normality appears to mix together statistical and prescriptive considerations. Knobe will discuss implications of these findings for a variety of questions in philosophy and cognitive science.