Abstract Background Because often introduced without proper validation studies, so-called “add-ons” to IVF have adversely affected in vitro fertilization (IVF) outcomes worldwide. All-freeze cycles (embryo banking, EB) with subsequently deferred thaw cycles are such an “add-on” and, because of greatly diverging reported outcomes, have become increasingly controversial. Based on “modeling” with selected patient populations, we in this study investigated whether reported outcome discrepancies may be the consequence of biased patient selection. Results In four distinct retrospective case control studies, we modeled in four cohort pairings how cryopreservation with subsequent thaw cycles affects outcomes differently in good-, average- and poor-prognosis patients: (i) 127 fresh vs. 193 frozen donor-recipient cycles to model best-prognosis patients; (ii) 741 autologous fresh non-donor IVF cycles vs. 217 autologous frozen non-donor IVF cycles to model average prognosis patients; (iii) 143 favorably selected autologous non-donor IVF cycles vs. the same 217 frozen autologous cycles non-donor to monitor good- vs. average-prognosis patients; and (iv) 598 average and poor-prognosis autologous non-donor cycles vs. the same 217 frozen autologous non-donor cycles to model poor vs. average prognosis patients. In best-prognosis patients, EB marginally improved IVF outcomes. In unselected patients, EB had no effects. In poor-prognosis patients, EB adversely affected IVF outcomes. Unexpectedly, the study also discovered independent-of-age-associated chromosomal abnormalities, a previously unreported effect of recipient age on miscarriage risk in donor-egg recipients. Conclusions In poor-prognosis patients, EB cycles should be considered contraindicated. In intermediate-prognosis patients EB does not appear to change outcomes, not warranting additional cost and time delays. Therefore, only good-prognosis patients are candidates for EB, though they will experience only marginal benefits that may not be cost-effective.