Two 3-bed 5-step and two 4-bed 7-step continuous feed pressure swing adsorption (PSA) cycle schedules were systematically devised and studied via simulation for producing high-purity O-2 (>= 99.5 vol %) at high recovery using a carbon molecular sieve (CMS). Two binary (O-2:Ar) and two ternary (O-2:Ar:N-2) feed mixtures were investigated to mimic the product produced from a typical O-2 PSA system. In addition to feed composition, the effects of light end (LE) versus heavy end (HE) equalization (Eq), two sequential Eq steps versus Eq followed by forced cocurrent depressurization (CoD)/light end pressurization (LEP) steps, feed flow rate (i.e., feed throughput theta), cycle time (t(c)), Eq step valve coefficient (C-v), and countercurrent depressurization (CnD) step C-v on the process performance were studied in terms of O-2 purity (Fur, vol %), O-2 recovery (Rec, %), and theta (L-STP, h/kg). In descending order, the best performance from each cycle was Pur = 99.5, Rec = 93.3, and theta = 455.7 for a 4-bed 7-step PSA cycle with LE Eq and forced CoD/LEP (M); Pur = 99.6, Rec = 85.7, and theta = 471.4 for a 4-bed 7-step cycle with two LE Eqs (II); Pur = 99.5, Rec = 77.4, and theta = 1089.5 for a 3-bed 5-step cycle with LE Eq (Ia); and Pur = 99.5, Rec = 68.7, and theta = 628.6 for a 3-bed 5-step cycle with HE Eq (Ib). The two C-v values exhibited only marginally unfavorable effects on Pur at smaller values. All four cycles were very sensitive to both t(c) and theta, with Pur decreasing and Rec increasing with decreasing theta, and Rec decreasing and Pur exhibiting a maximum with increasing t(c). For cycle III, the Pur exhibited a maximum with a change in the CoD end pressure, with more negative effects on Pur at longer t(c). Changing the levels of Ar or N-2 in the feed at fixed O-2 content had essentially no effect on Pur or Rec, while decreasing the level of O-2 in the feed had an almost proportionally negative effect on Pur and no effect on Rec. When N-2 was replaced with Ar in the feed, an unexpected equilibrium effect counteracted an expected kinetic effect, producing a counterintuitive negative result on Pur.