BACKGROUND: A thorough knowledge of sonography is essential in clinical practice. Therefore, sonography training is increasingly incorporated into the medical school curriculum, entailing different course models. The question arises which model is most effective to convey sustained sonographic skills. METHODS: Two different peer-assisted learning (PAL) sonography course models were developed as part of a clinical prospective study. The course content was based on the national resident curriculum of the German Society for Ultrasound in Medicine (DEGUM). Model A consists of a 10-week course and model B of a two-day compact course. Each model entailed 20 teaching units (TU). A script was used to prepare for each unit. Two modified OSCE exams of the ultrasound skills (max = 50 points per exam) were performed during the last teaching unit to assess the competence level. For subjective self-assessment and model evaluation, a questionnaire with a 7-point Likert scale was employed. RESULTS: A total of 888 students of the 3rd year participated as part of a voluntary elective in the study (744 in model A and 144 in model B). In the exams, participants in model A (median 43 points) scored significantly higher than those in model B (median 39; p < 0.01). Participants in model A (mean 1.71 points) obtained significantly higher mean competency gain scores in subject knowledge than model B (mean 1.43 points; p < 0.01) participants. All participants were satisfied with the course concept (A: mean 1.68 vs. B: mean 1.78 points; p = 0.05), the teaching materials (A: mean 1.81 vs. B: mean 1.69 points; p = 0.52), and the tutor's didactic skills (A: mean 1.24 vs. B: mean 1.15 points; p < 0.05). CONCLUSION: These results suggest that sonography-specific competency may be obtained through different course models, with a model stretching over several weeks leading to a higher competence level. Further research should assess the long-term retention of the skills obtained in different models.