ABSTRACT Objectives This study includes two separate parts: the objective for part A was to evaluate the practical manual therapy skills of undergraduate physiotherapy students who had learned manual therapy techniques either online or in classroom depending on the phases of the pandemic. The objective for part B was to evaluate in a randomized prospective design the effectiveness of video-based versus traditional teaching of a manual therapy technique. Design Cross-sectional cohort study (part A) and randomized controlled trial (part B) Setting University of Luebeck, undergraduate physiotherapy students in years 1–3 Method In part A, physiotherapy students who had learned manual therapy either online (during the pandemic) or in classroom (prior to and after the lock down periods of the pandemic) were videotaped while performing two manual techniques on the knee joint and on the lumbar spine. Recordings were analyzed independently by two blinded raters according to a 10-item list of criteria. Inter-rater reliability was assessed using Cohen’s kappa for each item. Performance across cohorts was analyzed using analysis of variance. In part B, students were randomized to learn a new technique on the cervical spine either from a lecturer or from the same lecturer on a video recording (independent variable). Practical performance of the technique was analyzed by two raters blinded to group allocation according to a 10-item list of criteria (dependent variable). Results were analyzed statistically by using ANCOVA with year of study as a covariate. Results Sixty-three and 56 students participated in part A and part B of the study, respectively. The inter-rater reliability for video analyses for both parts of the study was moderate (k = 0.402 to 0.441). In part A, there was no statistically significant difference across years of study for the practical performance of the technique on the back F(2,59) = 2.271; p = 0.112 or the knee joint F(2,59) = 3.028; p = 0.056. In part B, performance was significantly better when learned from a lecturer and practiced on a peer than when learned from a video and practiced on a rescue dummy (p < 0.001). Conclusion Practical skill performance can be acquired from videos but immediate skill reproduction is significantly better when the technique is presented by a lecturer in classroom and practiced on peer students.