The use of bibliometric indicators in research evaluation makes many hidden assumptions about scientists' publishing habits. This paper tests an assumption that scientists are reward oriented and attempt to publish in as prestigious channels as possible, seeking an optimal level in the hierarchy of publications. The data are based on interviews with teaching and research personnel in four university departments in the fields of zoology, biomedicine, and automation and control technology. The author concludes that in all studied fields scientists placed equal emphasis on the reward and communication functions of publishing. The actual publishing behaviour of biomedical scientists, nevertheless, accorded best with the assumption of seeking an optimal level of publishing in terms of prestige. By contrast, in zoology and technical fields, local and field-related publishing habits appeared strong.