Carbon-fiber microelectrodes are frequently used as chemical sensors in biological preparations. In this work, we evaluated the ability of cyclic voltammograms recorded at fast-scan rates to resolve neurochemicals when analyzed by principal component regression. A calibration set of 30 cyclic voltammograms was constructed from 9 different substances at a variety of concentrations. The set was reduced by principal component analysis, and it was found that 99.5% of the variance in the data could be captured with five principal components. This set was used to evaluate cyclic voltammograms obtained with one or two compounds present in solution. In most cases, satisfactory predictions of the identity and concentration of analytes were obtained. Chemical dynamics were also resolved from a set of fast-scan cyclic voltammograms obtained with the electrode implanted in a region of a brain slice that contains dopaminergic terminals. Following stimulation, principal component regression of the data resolved the changes in dopamine and pH that were evoked. In a second test of the method, vesicular release was measured from adrenal medullary cells and the data were evaluated with a calibration set composed of epinephrine and norepinephrine. Cells that secreted one or the other were identified. Overall, the results show that principal component regression with appropriate calibration data allows resolution of substances that give overlapping cyclic voltammograms.