Scientific careers are theoretically and practically important because they
link individuals with institutions as well as social structures with knowl
edge production. These mediating functions have to date not been systematic
ally dealt with. In this article, a neo institutionalist framework for the
analysis of careers in science is developed. Careers in science are treated
as products of overlapping institutions belonging to the different social
contexts in which scientists act simultaneously. These contexts (their spec
ialty, society and employment organization) yield specific institutions tha
t shape different work roles, which can be analytically distinguished. With
regard to a specialty's knowledge production, four different career stages
(apprentice, colleague, mentor and sponsor) can be distinguished on the ba
sis of dominant work roles. Society's institutions (e.g., language, educati
on and employment system) structure the international specialties, which ca
n be said to consist of national subsets. Organizations provide work roles
that integrate knowledge production into the employment system. Job sequenc
es become institutionalized as career lines that structure international in
ternal labour markets. While there is agreement in the literature that acad
emic organizations do not provide internal labour markets, so far it has be
en neglected that specialties have these properties. The stability of these
international internal labour markets currently seems to be endangered bec
ause both organizations and specialties can only balance the conflicting de
mands for fixed-term contracts and permanent positions when they grow. As a
conclusion of the theoretical discussion, a research program is outlined.