Two experiments are presented to explore the limits when matching a sample to a suspect utilising the hand as a novel biometric. The results of Experiment 1 revealed that novice participants were able to match hands at above-chance levels as viewpoint changed. Notably, a moderate change in viewpoint had no notable effect, but a more substantial change in viewpoint affected performance significantly. Importantly, the impact of viewpoint when matching hands was smaller than that when matching ears in a control condition. This was consistent with the suggestion that the flexibility of the hand may have minimised the negative impact of a sub-optimal view. The results of Experiment 2 confirmed that training via a 10-min expert video was sufficient to reduce the impact of viewpoint in the most difficult case but not to remove it entirely. The implications of these results were discussed in terms of the theoretical importance of function when considering the canonical view and in terms of the applied value of the hand as a reliable biometric across viewing conditions.