<p>Intravenous (IV) needle insertion is a crucial nursing skill that can be challenging to master. This study introduces a variability-adaptive IV needle insertion simulation integrating dual haptic feedback and mixed reality (MR) to create a realistic IV training environment. Unlike traditional training tools, our system incorporates graphic and haptic variable factors to simulate varying IV difficulty levels, reflecting real clinical conditions. Four variability-based difficulty levels, tuned and validated by nursing faculty, were designed and automatically rendered during training sessions to implement variability-adaptive training. To measure the effectiveness of this training, training impact study was conducted with 54 nursing students at Kent State University after the refinement from iterative usability studies(53 participants). The students were randomly assigned to one of two groups (variability and static). The performances were analyzed using objective measurements (completion time, insertion angles, success rate, and hand grip position), as well as a proposed objective performance evaluation metric, IV-score. The results showed learning improvement in both groups, with the variable group achieving higher learning gains (Variable: success rate + 22.4%, IV-score + 8.15; Static: success rate + 19.3%, IV-score + 6.6). These findings demonstrate the potential of variability-based training in our system as an early step towards achieving real clinic-like variability-adaptive training.</p> Graphical abstract <p></p>