The low rate of ongoing pregnancies in IVF cycles leads to a high number of repeated cycles in couples with previously failed attempts. Therefore it would be helpful to have a prediction about the chance of becoming pregnant in a repeated cycle. In a retrospective study the data of about 4246 cycles were analysed. Because the pregnancy rates in IVF- and ICSI cycles are generally different, these two groups were distinguished between and the outcome in patients with one, two or more attempts was analysed. The rate of ongoing pregnancies per patient was lower after IVF (24.9%) than after ICSI (32.9%), but was similar or even slightly increased in patients with more than one attempt. On the other hand, there was a high pregnancy rate with ICSI in the first two cycles (35.9%), but patients with more than two ICSI cycles had a significantly lower chance of becoming pregnant (20.7%). Factors that are known to influence the pregnancy rate, such as stimulation protocol, oocyte quality or number of transferred embryos, were similar in all groups. However, significantly reduced embryo quality with successive cycles was only observed in ICSI patients. There might be a negative selection of patients with poor embryo quality and previously failed attempts after ICSI, possibly due to an andrological factor. The differences between IVF- and ICSI patients are based on treatment indications, and andrological diseases are the predominant indication for ICSI. Although no correlation was found between changes in conventional sperm parameters and number of treated cycles, there might be a subgroup of andrological patients selected by repeatedly failed ICSI cycles. Chromosomal or genetic disturbances in spermatozoa of this subgroup could be the reason for failure.