BACKGROUND: Although intrauterine insemination (IUI) is one of the most common assisted reproductive technology methods in the world, the relative influence of various semen characteristics on the likelihood of a successful outcome is controversial. The aim of our study was to assess the results of IUI as a function of both the number of motile spermatozoa inseminated (NMSI) and the percentage of morphologically normal spermatozoa after preparation. METHODS: This was a retrospective study of 889 couples who underwent 2564 IUI cycles of ovarian stimulation with HMG or recombinant FSH in our centre between January 1991 and December 2000. RESULTS: A total of 331 clinical pregnancies were obtained, for a pregnancy rate/cycle of 12.91%. When the NMSI was <1 × 106, the pregnancy rate/cycle was significantly lower (3.13%) than in any of the subgroups with NMSI ≥2 × 106. Sperm morphology, assessed before or after preparation, was not in itself a significant factor that affected the likelihood of IUI success. Nonetheless, when the post-migration rate of normal sperm was <30%, the pregnancy rate/cycle was 5.43% when NMSI was <5 × 106 and 18.42% when NMSI was ≥5 × 106 (P=0.008). Pregnancy rates did not differ significantly according to NMSI when the percentage of normal sperm after preparation was ≥30%, or according to percentage of normal sperm when the NMSI was ≥5 × 106. CONCLUSIONS: Our results show that a minimum of 5 × 106 motile spermatozoa should be inseminated when the normal morphology of the sperm after preparation is <30%; the quantity compensates at least in part for the defective quality. If this threshold of NMSI cannot be obtained, IVF should be recommended.