The C, N, and S contents and δ13C and δ34S values were analyzed for 100 shale samples from ten formations, 3.0 to 2.1 Ga in age, in the central and eastern regions of the Kaapvaal Craton, South Africa. The Kaapvaal shales are characterized by generally low contents of organic C (range 0.06-2.79 wt%, average 0.47 wt%), N (range <0.01-0.09 wt%, average 0.1 wt%), and S (range <0.01-1.63 wt%, average 0.1 wt%). The low N/C (<0.005) and H/C (mostly ∼0.2) atomic ratios in kerogens from the shales indicated that the Kaapvaal shales lost considerable amounts of N, C, S, and H during diagenesis and regional metamorphism (up to the greenschist facies). From the theoretical relationships between the H/C ratios of kerogen and organic C contents of shales, the original C contents of the Archean and Proterozoic shales from the Kaapvaal Craton are estimated to be on average ∼2 wt%. These values are similar to the average organic C content of modern marine sediments. This suggests that the primary organic productivity and the preservation of organic matter in the ocean during the period of 3.0 to 2.1 Ga were similar to those in the Phanerozoic era, provided the flux of clastic sediments to the ocean was similar. This would also imply that the rate of O2 accumulation in the atmosphere-ocean system, which has equaled the burial rate of organic matter in sediments, has been the same since ∼3.0 Ga. The δ34S values of bulk-rock sulfides (mostly pyrite) range from +2.7 to +7.4‰ for seven sulflde-rich samples of ∼2.9 Ga to ∼2.6 Ga. These values are consistent with a suggestion by Ohmoto (1992) and Ohmoto et al. (1993) that most pyrite crystals in Archean shales were formed by bacterial reduction of seawater sulfate with δ34S values between +2 and +10‰, and that the Archean seawater was sulfate rich. Changes in the δ13Corg values during maturation of kerogen were evaluated with theoretical calculations from the experimental data of Peters et al. (1981) and Lewan (1983), and from the observations by Simoneit et al. (1981) on natural samples. These evaluations suggest that the magnitudes of δ13Corg increase are much less than those estimated by Hayes et al.(1983) and Des Marais et al. (1992), and only about 2 to 3‰ for the kerogens that decreased their H/C ratios from 1.5 to less than 0.3. Based on the relationships among sulfide-S contents, organic-C contents, and δ13Corg values, four different types of depositional environments are identified for the Archean and early Proterozoic shales in the Kaapvaal Craton: (I) euxinic marine basins, characterized by normal marine organisms with δ13Corg = -33 ± 3‰ ; (II) near-shore, oxic marine environment, characterized by normal marine organisms with δ13Corg = -31 ± 3‰; (III) hypersaline, low-sulfate lakes, characterized by organisms with δ13Corg = -26 ± 3‰; and (IV) euxinic, marine basins which supported the activity of methanogenic and methanotrophic bacteria and accumulated organic matter with δ13Corg = -43 ± 3‰. In contrast to the currently popular model positing a global anoxic ocean prior to ∼2.2 Ga (e.g., Des Marais et al, 1992; Hayes, 1994; Logan et al., 1995), this study suggests that the development of anoxic basins, which accumulated Group II and IV sediments, occurred only regionally and episodically during the period between 3.0 Ga and 2.1 Ga. This further suggests that the normal ocean has been oxic since at least ∼3.0 Ga. Diversifications of environments, as well as of biological species, had already occurred ∼3.0 Ga. The carbon isotope mass balance calculation suggests that the removal rates of organic C and carbonate C from the ocean and the weathering rates of organic C and carbonate C on the continents during the 3.0-2.1 Ga period were basically the same as those in the Phanerozoic era. This would have been possible only if the atmospheric PO2 level had been basically constant since at least 3.0 Ga. The results of this study, therefore, add to a growing list of evidence that the atmosphere has been oxic (i.e., PO2 >1%PAL) since at least 3.0 Ga. The list of evidence includes the sulfur isotope data on Archean sedimentary rocks (Ohmoto and Felder, 1987; Ohmoto et al., 1993), the Fe3+/Ti ratios of paleosols (Ohmoto, 1996), and the paragenesis of minerals in the "detrital" gold-uranium ores in pre-2.0 Ga quartz pebble beds that suggests nondetrital origins for uraninite and pyrite in these deposits (Baraicoat et al., 1997).