In our medical school, we changed from a lecture-based method to a team-based learning (TBL) method to teach "polyneuropathies" in the neurology clerkship starting from the 2014 to 2015 academic year. Real patients were used instead of written scenarios in TBL sessions. This study aimed to compare former lecture-based and the current TBL methods in terms of knowledge retention, in-class learner engagement, and learner reactions. First, we determined in-class engagement and satisfaction of the students for the lectures given in the 2013-2014 academic year. The following year, besides the same criteria, we also determined individual (IRAT) and group readiness test (GRAT) scores in the TBL group. End-of-clerkship exam scores for both groups were recorded. Additionally, opinions of patients about their experiences throughout the TBL process were determined. One year later (2015 for lecture and 2016 for TBL), both groups sat for an MCQ test to determine their knowledge retention levels. We found no difference between groups regarding end-of-clerkship exam scores. The mean knowledge retention test score of the TBL group was significantly higher than that of the lecture group (5.85 ± 1.74 vs. 3.28 ± 1.70). The differences between IRAT, GRAT, and retention test scores in the TBL group were significant. The mean student satisfaction score on a five-point scale was 3.01 ± 0.9 (median = 3) in the lecture group and 4.11 ± 1.1 (median = 4) in the TBL group. Our results seem encouraging for use of TBL performed with real patients in neurology education to achieve better long-term knowledge retention and higher in-class engagement and student satisfaction.