Human embryo culture under 2–8% O2 is recommended by ESHRE revised guidelines for good practices in IVF labs. Nevertheless, notably due to the higher costs of embryo culture under hypoxia, some laboratories perform embryo culture under atmospheric O2 tension (around 20%). Furthermore, recent meta-analyses concluded with low evidence to a superiority of hypoxia on IVF/ICSI outcomes. Interestingly, a study on mice embryos suggested that oxidative stress (OS) might only have an adverse impact on embryos at cleavage stage. Hence, we aimed to demonstrate for the first time in human embryos that OS has a negative impact only at cleavage stage and that sequential culture conditions (5% O2 from Day 0 to Day 2/3, then «conventional» conditions at 20% O2 until blastocyst stage) might be a valuable option for human embryo culture. 773 IVF/ICSI cycles were included in this randomized clinical trial from January 2016 to April 2018. At Day 0 (D0), patients were randomized using a 1:2 allocation ratio between group A (20% O2; n = 265) and group B (5% O2; n = 508). Extended culture (EC) was performed when ≥ 5 Day 2-good-quality-embryos were available (n = 88 in group A (20% O2)). In subgroup B, 195 EC cycles were randomized again at Day 2 (using 1:1 ratio) into groups B’ (5% O2 until Day 6 (n = 101)) or C (switch to 20% O2 from Day 2 to Day 6 (n = 94). Fertilization rate, cleavage-stage quality Day 2-top-quality-embryo (D2-TQE), blastocyst quality (Day 5-top-quality-blastocyst (D5-TQB) and implantation rate (IR) were compared between groups A and B (= cleavage-stage analysis), or A(20% O2), B’(5% O2) and C(5%-to-20% O2). Overall, characteristics were similar between groups A and B. Significantly higher rates of early-cleaved embryos, top-quality and good-quality embryos on Day 2 were obtained in group B compared to group A (P < 0.05). This association between oxygen tension and embryo quality at D2 was confirmed using an adjusted model (P < 0.05). Regarding blastocyst quality, culture under 20% O2 from Day 0 to Day 6 (group A) resulted in significantly lower Day 5-TQB number and rates (P < 0.05) compared to both groups B’ and C. Furthermore, blastocyst quality was statistically equivalent between groups B’ and C (P = 0.45). At Day 6, TQB numbers and rates were also significantly higher in groups B’ and C compared to group A (P < 0.05). These results were confirmed analyzing adjusted mean differences for number of Day 5 and Day 6 top quality embryos obtained in group A when compared to those respectively in groups B’ and C (P < 0.05). No difference in clinical outcomes following blastocyst transfers was observed. These results would encourage to systematically culture embryos under hypoxia at least during early development stages, since OS might be detrimental exclusively before embryonic genome activation.