In March 1994, we used the newly refurbished Hubble Space Telescope (HST) to obtain global imaging of Io at five wavelengths between 0.34 and 1.02 μm, with a spatial resolution of 160 km. The images provided the clearest view of Io since Voyager and the first systematic observations in the wavelength range 0.7-1.0 μm. We have produced absolutely calibrated global mosaics of Io's reflectance in all our five wavelengths. The near-infrared images reveal that the 0.55- to 0.7-μm absorption edge seen in Io's disk-integrated spectrum has a very different spatial distribution from the better-known 0.40- to 0.50-μm absorption edge studied by Voyager, and must be generated by a different chemical species. The 0.55- to 0.7-μm absorption edge is strongly concentrated in the pyroclastic ejecta blanket of the volcano Pele, at a few much smaller discrete spots, and probably also in the polar regions. The Pele ejecta spectrum is consistent with the idea that S 2O, partially decomposed to S 4(and probably S 3), may be the species responsible for the 0.55- to 0.7-μm absorption edge at Pele and elsewhere on Io, though S 4generated by other processes may also be a possibility. S 2O can be produced by high-temperature decomposition of SO 2gas, and the high temperature of the Pele volcano may account for its concentration there. Spectral anomalies of comparable size and prominence are not seen around the other "Pele-type" volcanos Surt and Aten (A. S. McEwen and L. A. Soderblom, 1983, Icarus55, 191-217), suggesting that these volcanos, if chemically similar to Pele, are much less active. The spectrum of high-latitude regions is similar to that of quenched red sulfur glass, and if this similarity is not coincidental, the glass may be preserved here by the low polar surface temperatures. Alternatively, the low polar temperatures may preserve sulfur that has been reddened by radiation. There are many changes in albedo patterns in the 15 years between Voyager and these HST observations, but these are generally subtle at HST resolution and are not strongly concentrated in longitude; however there was a major brightening of a 400-km-diameter region centered on Ra Patera between March 1994 and repeat HST observations in July 1995, which was a larger albedo change than any seen in the previous 15 years. This was presumably due to a large eruption at Ra Patera, as confirmed by Galileo images. Long-exposure eclipse images of Io at 1.02 μm on March 6, 1994, place strong limits on the area of exposed silicate magma on Io at the time of the observations.