[Abridged] We present reliable multiwavelength identifications and high-quality photometric redshifts for the 462 X-ray sources in the ~2 Ms Chandra Deep Field-South. Source identifications are carried out using deep optical-to-radio multiwavelength catalogs, and are then combined to create lists of primary and secondary counterparts for the X-ray sources. We identified reliable counterparts for 446 (96.5%) of the X-ray sources, with an expected false-match probability of ~6.2%. A likelihood-ratio method is used for source matching, which effectively reduces the false-match probability at faint magnitudes compared to a simple error-circle matching method. We construct a master photometric catalog for the identified X-ray sources including up to 42 bands of UV-to-infrared data, and then calculate their photometric redshifts (photo-z's). The reliability of the photo-z's is evaluated using the subsample of 220 sources with secure spectroscopic redshifts. We achieve an accuracy of ~1% [|Delta z|/(1+z)] and an outlier fraction of ~1.4% for sources with spectroscopic redshifts. We performed blind tests to derive a more realistic estimate of the photo-z quality for sources without spectroscopic redshifts. We expect there are ~9% outliers for the relatively brighter sources (R<26), and the outlier fraction will increase to ~15-25% for the fainter sources (R>26). The typical photo-z accuracy is ~6-7%. The outlier fraction and photo-z accuracy do not appear to have a redshift dependence (for z~0-4). These photo-z's appear to be the best obtained so far for faint X-ray sources, and they have been significantly (>50%) improved compared to previous estimates of the photo-z's for the X-ray sources in the ~2 Ms Chandra Deep Field-North and ~1 Ms Chandra Deep Field-South.