Commissioned by the Bank of Canada to help improve the detection of counterfeit currency, we designed a series of tests of performance to explore the contributions of note quality, sensory modality, training, security features and demographic variables to the accuracy of counterfeit detection with three different note types. In each test, participants (general public, and cash handlers, divided amongst commercial cash handlers and bank tellers) were presented with notes, one at a time, for up to seven seconds, and were asked to judge whether each note was genuine or counterfeit. With whole note inspection, overall accuracy was about 80%. When the security features were tested individually, the Optical Security Device (OSD) was the best feature, the hidden number was the worst, and the portrait, maple leaves, fluorescence, and microprinting were intermediate. Accuracy was higher with notes that could be seen but not touched than vice versa. Cash handlers were 74% correct with touch alone and adding touch to vision significantly improved counterfeit detection. This paper will demonstrate how performance differences between the different note types can be explained in terms of the efficacy of the individual security features incorporated into the notes.