Photoluminescence spectra of single-walled carbon nanotubes (SWCNTs) have been recorded and analyzed for selected individual nanotubes and structurally sorted bulk samples to clarify the nature of secondary emission features. Room temperature spectra show, in addition to the main peak arising from the E 11 bright exciton, three features at lower frequency, which are identified here (in descending order of energy difference from E 11 emission) as G 1 , X 1 , and Y 1 . The weakest (G 1 ) is interpreted as a vibrational satellite of E 11 involving excitation of the ∼1600 cm −1 G mode. The X 1 feature, although more intense than G 1 , has a peak amplitude only ∼3% of E 11 . X 1 emission was found to be polarized parallel to E 11 and to be separated from that peak by 1068 cm −1 in SWCNTs with natural isotopic abundance. The separation remained unchanged for several (n,m) species, different nanotube environments, and various levels of induced axial strain. In 13 C SWCNTs, the spectral separation decreased to 1023 cm −1 . The measured isotopic shift points to a phonon-assisted transition that excites the D vibration. This supports prior interpretations of the X 1 band as emission from the dark K-momentum exciton, whose energy we find to be ∼230 cm −1 above E 11 . The remaining sideband, Y 1 , is red-shifted ∼300 cm −1 from E 11 and varies in relative intensity among and within individual SWCNTs. We assign it as defect-induced emission, either from an extrinsic state or from a brightened triplet state. In contrast to single-nanotube spectra, bulk samples show asymmetric zero-phonon E 11 peaks, with widths inversely related to SWCNT diameter. An empirical expression for this dependence is presented to aid the simulation of overlapped emission spectra during quantitative fluorimetric analysis of bulk SWCNT samples.