With more advanced features loaded, smartphones nowadays are used not only for telecommunication but also for many emerging applications, such as m-banking. In this paper, we propose a novel non-intrusive authentication mechanism using the information collected from the orientation sensor of the smartphone. This new approach is based on the hypothesis that a user has a unique way to hold and operate his/her smartphone while working on some apps; and such behavioral biometrics can be captured from the readings of the orientation sensor. We design an authentication mechanism that adopts 53 new features transformed from those readings. To validate this hypothesis, we have developed an application to collect user's behavioral biometrics of up-down flicks and left-right flicks from the orientation sensor. The experimental results show that the proposed approach has an equal error rate about 6.85%. We find that the feature subset selected to build an authentication model with satisfactory performance is generally small, varying 3 to 8 for different users. We also find that the feature subsets are significantly different among different users. Finally, we show that the proposed non-intrusive mechanism can be used together with existing intrusive mechanisms, such as password and/or fingerprints, to build a more robust authentication framework for smartphone users.