INTRODUCTION: References are fundamental in our studies; an adequate selection is asimportant as an adequate description. OBJECTIVES: To determine the number of errors in a sample of references found in Peruvian medical journals. METHODS: We reviewed 515 scientific papers references selected by systematic randomized sampling and corroborated reference information with the original document or its citation in Pubmed, LILACS or SciELO-Peru. RESULTS: We found errors in 47,6% (245) of the references, identifying 372 types of errors; the most frequent were errors in presentation style (120), authorship (100) and title (100), mainly due to spelling mistakes (91). CONCLUSIONS: References error percentage was high, varied and multiple. We suggest systematic revision of references in the editorial process as well as to extend the discussion on this theme. KEY WORDS: references, periodicals, research, bibliometrics.