Background Assisted reproductive treatment for women with endometriosis and diminished ovarian reserve (DOR) is challenging, and the optimal controlled ovarian stimulation (COS) protocol for this population remains uncertain. This study compared four commonly used COS regimens for IVF/ICSI in women with EM and DOR. Methods In this retrospective cohort (2014-2023), 859 IVF/ICSI cycles in women aged 40 years or younger with EM and DOR were allocated to four COS protocols (ultra-long, long, short, antagonist). The primary outcomes were cumulative live birth rate (CLBR) and live birth rate (LBR) per started cycle; clinical pregnancy rate (CPR) was secondary. Between-protocol differences were assessed using multivariable logistic regression, adjusting for age, infertility duration, history of endometrioma cystectomy, basal FSH, LH, E2, AMH, antral follicle count, BMI, and COS protocol, with additional stabilized inverse-probability-weighting and 1:1:1:1 propensity score matching used to address confounding by indication. Results The analysis included a total of 859 cycles, which were allocated into four treatment groups: the short protocol (n=295), the long protocol (n=145), the ultra-long protocol (n=202), and the antagonist protocol (n=217). Among these, the ultra-long protocol yielded the highest CLBR of 48.5%, LBR of 36.6%, and CPR of 41.1%. After PSM, the advantage difference of the ultra-long protocol is more significant. A subsequent multivariable logistic regression analysis confirmed that the ultra-long protocol was significantly more effective for achieving a live birth than the other protocols in the study: long protocol (OR: 0.577, P = 0.042), short protocol (OR: 0.608, P = 0.028), and antagonist protocol (OR: 0.486, P = 0.004). Similarly, the short (OR: 0.559, P = 0.010) and antagonist (OR: 0.484, P = 0.004) protocols were less effective for CPR. Conclusion For infertile women aged 40 years or younger with endometriosis and DOR undergoing fresh embryo transfers, the ultra-long protocol appears to be more beneficial than the long, short, and antagonist protocols in terms of cumulative live birth, live birth, and clinical pregnancy rates.