Traditional learning and assessment systems are overwhelmed when it comes to addressing the complex and multi-dimensional problems of clinical communication and professional practice. This paper shows results of a training program in clinical communication under Problem Based Learning (PBL) methodology and correlation between student self-assessment and teachers' assessment. This involves a qualitative-quantitative cross-sectional study in usual practice in the 2nd year of the degree in Medicine. Teaching methodology is PBL, including 15 associate professors and 90 students. Educational tools for learning: PBL cases and seminars (video recorded, theoretical-practical lectures). Assessment tools: Tutorials on those cases worked on PBL (40%), knowledge test (30%), assessment of a case with PBL methodology (20%) and video recording report (10%). Communication skills are evidenced by CICCA-D scale (Connect-Understand-Identify-Agree-Help-Decision). Variables: academic performance, score on CICCA-D and academic methodological assessment. The analysis is carried out using descriptive statistics, calculating the intra-class correlation coefficients and weighted Kappa index with quadratic weights. 92.2% of students passed the course on the first round. In a range between 0 and 34 points students' self-assessment scored 13 (SD ± 5) points and teachers' 16 (SD ±7). A weak (21%-41%) or poor (< 20%) correlation was obtained between teachers and students for all questions on CICCA-D. The authors suggest a summative assessment using different instruments and techniques to assess clinical communication skills from the first year onwards, and highlight the key role of self-assessment, peer assessment and the use of video recording techniques along with feedback in formative assessment.