Purpose: This study aimed to evaluate the effectiveness of using a virtual patient (VP) software program in increasing clinical reasoning skills confidence among medical students. Design/methodology/approach: A quasi-experimental design was used to compare the outcomes of students' confidence levels on clinical reasoning between problem-based learning (PBL) and VP program with 122 pre-clinical Year 2 medical students. History taking, physical examination, clinical investigations and diagnosis was investigated using four PBL cases in the oncology block. Pre- and post-differences in mean confidence scores between the arms were compared, as well as mean difference by type of case and skill category. Findings: A total of 122/156 (78.2%) students participated; n = 55 were from 7 VP groups and n = 67 were from 13 control arm groups. For the primary outcome, the VP arm showed a statistically significant increase in confidence measures among 11/16 (4 cases × 4 skills) categories, compared to 4/16 for the control arm. The proportion of the students who indicated an improved confidence was statistically significant for the cervical cancer case for physical examination (60.0% improved in VP vs 12.5% in control), investigation (60.0% VP vs 18.8% control) and diagnosis (60.0% in VP vs 25.0% in control). Finally, analysis by case showed an increment in overall mean scores from the start to end of the case within the VP arm while the pattern was erratic in the control arm cases. Originality/value: The study results showed that incorporating the VP into PBL was more effective in increasing students' clinical reasoning confidence levels compared to the usual PBL. As the study utilized existing PBL cases, it demonstrates how medical schools can incorporate digital VP tools into pre-clinical years before students' transition to learning from actual patients in the clinical years.