Generally speaking, the three‐year synchronous impact factor is larger than the two‐year one. This follows from theoretical models derived from observations based on ISIÒs database. In this article we present an exception to this general rule, based on data from the Chinese Science Citation Database (CSCD). In 1998 42% of this databaseÒs source journals did not follow the expected trend. As a possible explanation we note that, contrary to intuition, in the CSCD the changes in the number of both publications and citations are largely independent. It is, however, not ruled out that the observed discrepancies are nothing but statistical fluctuations of the basic publication‐citation model.