This paper is a diachronic analysis of a corpus of 180 titles drawn from Case Reports (CRs) published in the BMJ and BMJ Case Reports between 1840 and 2009. The corpus was divided into three blocks, and the frequency of occurrence of 69 text-internal variables was recorded in each title. Between-block comparisons were carried out, and Student's t-tests were applied to the quantitative results. Our findings show that CR titles have evolved over the 160-year period studied in the sense that they have increased in length, syntactic complexity, semantic richness and title-type diversity. Authorship patterns and collaboration practices have changed too. Although internationalization of case reporting has increased over time, today's preferred practice is still local collaboration. The only variable that has remained constant over the years is the nominal nature of CR titles. We put forth several social and scientific factors that could account for the various shifts observed. We claim that non-informativeness of CR titles that persisted over time can be explained by the fact that CR authors are reluctant to give a generalization flavor to their findings.