Homologous chromosome segregation errors during meiosis I are common and generate aneuploid embryos. Here, we provide a reason for this susceptibility to mis-segregation by live cell imaging of mouse oocytes. Our results show that stable kinetochore-microtubule attachments form in mid-prometaphase, 3-4 hours before anaphase. This coincided with the loss of Mad2 from kinetochores and with the start of anaphase-promoting complex/cyclosome (APC/C)-mediated cyclin B1 destruction. Therefore, the spindle assembly checkpoint (SAC) ceased to inhibit the APC/C from mid-prometaphase. This timing did not coincide with bivalent congression in one-third of all oocytes examined. Non-aligned bivalents were weakly positive for Mad2, under less tension than congressed bivalents and, by live-cell imaging, appeared to be in the process of establishing correct bi-orientation. The time from when the APC/C became active until anaphase onset was affected by the rate of loss of CDK1 activity, rather than by these non-aligned bivalents, which occasionally persisted until anaphase, resulting in homolog non-disjunction. We conclude that, in oocytes, a few erroneous attachments of bivalent kinetochores to microtubules do not generate a sufficient SAC 'wait anaphase' signal to inhibit the APC/C.