In this study, we propose the possibility of generating several parallel and independent random bitstreams from the time-varying optical emission spectra of an atmospheric pressure air microplasma system. This is achieved by splitting the plasma arc emission into discrete wavelengths using an optical spectrometer and then monitoring the fluctuating intensities of each wavelength as an independent time series. As a proof of concept, we considered eight wavelengths centered at 377.8, 389.1, 425.8, 591.4, 630.5, 673.0, 714.2, and 776.4 nm corresponding to atomic emissions lines from species either from the surrounding atmospheric air gap or from the electrodes' materials. NIST SP 800-22 statistical randomness tests and other statistical estimates (auto- and cross-correlation analysis and binary vector similarity measures) are subsequently applied to the binarized data, and the obtained results confirm the possibility of generating several parallel and independent random bitstreams from the microplasma system. The data throughput is, however, relatively low with the optical setup we used, which can be improved using faster spectrometry.