Introduction The coronavirus 2019 pandemic highlighted virtual learning (VL) as a promising tool for medical education, yet its effectiveness in teaching clinical reasoning (CR) remains underexplored. Past studies have suggested VL can effectively prepare students for clinical settings. Informed by the Milestones of Observable Behaviours for CR (MOBCR) and whole-case theoretical frameworks, the Mock Wards (MW) program was created using a novel blended in-person learning (IPL) and VL platform. MW consisted of case-based small-group formats for medical students interested in learning approaches and differentials to commonly encountered presenting symptoms and diagnoses in internal medicine. This study sought to use MW's blended design to qualitatively analyze CR development and compare its utility between VL and IPL. Methods Qualitative analysis was conducted using in-depth semi-structured interviews with first-year pre-clerkship medical students (n = 8) who completed the MW program and participated in the study. The interview guide was informed by the MOBCR framework. Interview transcripts were analyzed using a directed qualitative content analysis approach. Translational coding and HyperRESEARCHTM (Researchware, Inc., Randolph, MA) software-generated mind maps guided the theme development. Results Three overarching themes were constructed: (1) tailoring pedagogical frameworks to learning modalities, (2) learning through interactivity, and (3) balancing accessibility with learner engagement. Participants emphasized that teaching CR skills is modality-specific and not fully interchangeable, with IPL being superior in facilitating social cohesion, non-verbal communication, and feedback. In contrast, VL required structured approaches and relied more on verbal communication and pre-made digital materials. IPL also enhanced interactivity, peer relationships, and spontaneous communication, whereas VL faced challenges such as social awkwardness and technological constraints hindering effective collaboration. VL provided superior accessibility to facilitate distributed learning and management of concurrent academic obligations. Conclusion The MW-blended platform highlights the importance of focusing on modality-tailored pedagogies, emphasizing group interactability, and balancing VL accessibility against decreased engagement within the IPL environment when teaching CR skills. Blended education models may benefit from a scaffolding approach, using IPL as a prerequisite to VL to improve CR development and alignment within a learner's zone of proximal development.