Creating protocols (or guidelines, standards) for medical work is becoming big business. The pros and cons of protocols have been reiterated many times: advocates argue that protocols may enhance the quality of care, reduce unwanted variations in practice, and may help to render medical practice more scientific. Critics, on the contrary, argue that protocols will lead to cookbook medicine, to de-skilling, and to a reduced quality of care. In the continuing reiteration of these claims, they have more and more become removed from the actual practices of medical work and of the creation and use of protocols. Building on empirical research of these practices, this paper attempts to revitalise the debate. Four problems of the protocol are discussed: the protocol reinforces the tendency to perceive and describe medical action as an individual, formally rational process; the protocol contributes to the widespread illusion of the single answer; the protocol contributes to the loss of importance of information and interventions which are difficult to explicate and/or to quantify; and the protocol will lead to an increasing bureaucratisation and regulation of health care practices. Judicious use of protocols has several promises, which are discussed subsequently: explicating a scheme of actions creates a forum for discussion, and delegating parts of the decision-making process to a protocol may lead to personnel acquiring new competences.