The effects of different artificial insemination (AI) techniques and sperm doses on pregnancy rates of normal Hanoverian breed mares and mares with a history of barrenness or pregnancy failure using fresh or frozen-thawed sperm were investigated. The material included 187 normal mares (148 foaling and 39 young maiden mares) and 85 problem mares with abnormal reproductive history. Mares were randomly allotted into groups with respect to AI technique (routine AI into the uterine body, transrectally controlled deep intracornual AI ipsilateral to the preovulatory follicle, or hysteroscopic AI onto the uterotubal junction ipsilateral to the preovulatory follicle), storage method of semen (fresh, frozen-thawed), AI volume (0.5, 2, 12 ml), and sperm dose (50 x 10(6) or 300 x 10(6) progressively motile sperm (pms) for fresh semen and 100 or 800 x 10(6) frozen-thawed sperm with >35% post-thaw motility). The mares were inseminated once per cycle, 24 h after hCG administration when fresh semen was used, or 30 h for frozen-thawed semen. Differences in pregnancy rates between treatment groups were analyzed by Chi-squared test, and for most relevant factors (insemination technique, mare, semen, and stallion) expectation values and confidence intervals were calculated using multivariate logistic models. Neither insemination technique, volume, sperm dose, nor mare or stallion had significant effects (P > 0.05) on fertility. Type of semen, breeding mares during foal heat, and an interaction between insemination technique, semen parameters, and mares did have significant effects (P < 0.05). In problem mares, frozen semen AI yielded significantly lower pregnancy rates than fresh semen AI (16/43, 37.2% versus 25/42, 59.5%), but this was not the case in normal mares. In normal mares, hysteroscopic AI with fresh semen gave significantly (P < 0.05) better pregnancy rates than uterine body AI (27/38, 71% versus 18/38, 47.3%), whereas in problem mares this resulted in significantly lower pregnancy rates than uterine body AI (5/15, 33.3% versus 16/19, 84.2%). Our results demonstrate that for problem mares, conventional insemination into the uterine body appears to be superior to hysteroscopic insemination and in normal mares, the highest pregnancy rates can be expected by hysteroscopic insemination.