This paper proposes an automatic approach that detects students' learning styles in order to provide adaptive courses in Moodle. This approach is based on students' responses to the ILS and the analysis of their interaction behavior within Moodle by applying a data mining technique. In conjunction to this, an adaptive mechanism that was implemented in Moodle is presented. This adaptive mechanism builds the user model based mainly on the proposed approach for automatic detection of learning styles in order to adapt the presentation and the proposed navigation to students' different learning styles and educational objectives. An evaluation study was conducted to evaluate the proposed approach for automatic detection of learning styles and the effect of the adaptive mechanism. Two groups of students were formed, namely the experimental and the control. The first had access to a Moodle course that automatically detected their learning styles and exploited the adaptive mechanism, whilst the second had access to the standard version of a Moodle course. The results were promising since they indicated that our proposed approach for automatic detection of learning styles attained adequate precision compared to other works, even though the patterns considered are less complex. Additionally, the results indicated that the adaptive mechanism positively affected students' motivation and performance.