In 1987, a Letter to the Editor published in Cell (1) outlined the need of a careful use of the term “homology” in literature, being wrongly used instead of “similarity” in articles describing comparison of protein and nucleic acid sequences. Three different types of mistakes were evidenced with the related misleading consequences: i) the observed similarity, indicated as “homology”, was surprisingly commented as non related to evolutionary homology; ii) the observed similarity, again erroneously called “homology”, was not accompanied by any consideration about the evolutionary relationships, and the reader was erroneously induced to consider the sequences as homologues; and iii) the found similarities, called “homologies”, were used to support the hypothesis of evolutionary homology. To prevent these misinterpretations and their consequences, the Authors of the Letter recommended to use the term “homology” only in its precise meaning of “having a common evolutionary origin”. What did the community learn from the lesson of twenty years ago? We searched the PubMed archive for articles published in 2007, having the keyword “homology” in its abstract or title. By excluding the cases in which “homology” is part of a gene or protein name (as an example, Bcl-2 homology domain) or indicates a procedure (e.g. “homology modelling”) we finally selected 1966 abstracts. The term was correctly used in 57% of those abstracts, while in the remaining 43% the term was still not used correctly. In 264 abstracts, homology was associated to a percentage of similarity, and in 452 abstracts it was used together with terms as “high”, “low” and so on. In 94 abstracts the term “homology” was defined “significant” and finally, another 28 abstracts reported a misleading association indicating somehow a quantitative evaluation of the homology. A similar search in the abstracts of articles published in 1986, one year before the publication of that Letter, found that the term ‘homology’ was incorrectly used in 51% of abstracts analyzed. Looking at the different types of errors, we noticed that the frequency of the expression ‘percentage of homology’ was more or less the same as in 2007, whereas the frequency of expressions such as ‘high homology’, ‘low homology’ and so on, was higher in 1986 than it was in 2007. This simple search with PubMed does not include the whole text of the articles, so we cannot exclude that in the article the Authors use “homology” correctly, as well as that many other errors appear in the text, not in the abstract. We report also the 20 journals in which the abstracts containing ‘homology’ appeared more frequently in 1986, and compared with the data in 2007. Finally, an investigation was also performed about the language in the whole article was written. We can simply conclude that, despite a little improvement in the term usage after 20 years, bad habit die hard, and that researchers should always read and learn from the past lessons.