Can research be studied in a way that is neither logical reconstruction nor empirical psychology or sociology of science? In contemporary philosophy of science this is usually denied—in spite of the recent 'paradigm shift' there. A system-philosophy approach in theory of research is outlined by means of some models : a research enterprise is viewed as a productive, innovative system, the research process as a transformation of complexes of knowledge-problems-instruments (software and hard ware). The direction this development takes is guided (...) by preconceptions about the subject matter and a programmatic conception of the discipline based on them ('internal steering factors'). The dynamics of the research process are schematized as a co-agency of 'theoretical and 'empirical' moments, which is viewed as a flow of problems, information, and conceptual frameworks. Empirically discovered pieces of knowledge may resist all attempts to explain them: the flow is disturbed, and only an extraordinary innovation on the 'theoretical' level can restore the balance. This involves a shift in perspective: a shift in the 'internal steering factors'. (In this way one type of 'scientific revolution' (Kuhn's type) is conceptualized). After a perspectival shift one will ask 'Does it constitute progress ?' Types of criteria are suggested. Although they do not apply to all research situations, they over-arch research-traditions. Eventually the above models are applied in the reflection of Research Theory on itself. Research Theory should improve our knowledge about knowledge-production. This knowledge should contribute to improving our image of science and the researcher's sensitivity—by providing better tools for concept ualizing research situations and for imagining possible alternatives. (shrink)