Introduction: One of the many notions that have been introduced in education in the last 50 years is that of learning styles. This idea is very popular and enjoys good acceptance. However, from a scientific perspective this notion is very controversial. In this sense, Coffield, Moseley, Hall, and Ecclestone, made the first systematic review on the subject, and showed that the psychometric instruments of learning styles do not have adequate levels of reliability and validity. However, the results of their review have not been updated since 2004. This contribution presents a systematic review of these aspects for the 2005-2010 period, with the aim of updating the information on the subject. Method: Fifty-eight documents were examined, which contained enough information about the reliability and validity of the following inventories: Kolb's LSI, Felder's ILS, Sternberg's TSI, the VAK-VAKT-VARK sensory inventories, the inventories based on the model proposed by Entwistle (ASSIST, LASSI and RASI), Biggs' SPQ, Honey and Mumford's LSQ and Gregorc's Style Delineator (GSD). Results: The analysis of the psychometric properties indicates that: (1) more than half of the inventories surveyed lack internal consistency and predictive validity, (2) around 31% have no test-retest reliability, (3) there is is no evidence on a at least one reliability and validity criteria for close to 31 % of them (4) Jackson's Learning Styles Profiler (LSP) remains an inventory without evidence on all reliability and validity criteria measures. Discussion and Conclusion: Despite the popularity and acceptance of learning styles, the initial recommendation made 14 years ago by Coffield, Moseley, Hall, and Ecclestone, of not basing pedagogical interventions solely on any of the learning styles instruments is still valid.