We compared conventional methods for laparoscopic and cervical artificial insemination (AI) to a transcervical AI procedure (Guelph System for Transcervical AI; GST-AI) for use with frozen semen in Merino ewes. The GST-AI procedure was performed by an experienced operator in Experiment 1 (771 ewes) and by 2 inexperienced operators in Experiment 2 (555 ewes). In Experiment 1, intrauterine insemination by GST-AI was achieved in 76% of the ewes. The pregnancy rate at Day 70 for ewes inseminated by laparoscopy (48%, 120 251 ) was higher (P<0.01) than for ewes inseminated by either intrauterine GST-AI (32%, 64 201 ) or cervical AI (9%, 24 256 ). The overall (intrauterine and intracervical) pregnancy rate for GST-AI was 26% (68 264 ) and was unaffected by depth of insemination within the cervix. Pregnancy rates were unaffected by ram or day of insemination. In Experiment 2, the operators achieved intrauterine inseminations by GST-AI in 43% (78 182 ) of the ewes, with a significant operator effect (P<0.01) on depth of cervical penetration. The pregnancy rate to intrauterine GST-AI (40%, 31 78 ) did not differ from that to laparoscopic insemination. The total pregnancy rate for GST-AI in Experiment 2 (19%, 34 182 ) was lower (P<0.05) than that for laparoscopic AI (39%, 72 187 ) but superior (P<0.05) to that for cervical AI (1%, 1 186 ). The GST-AI pregnancy rates were affected by depth of AI (P<0.01) and by operator (P<0.05). It is concluded that GST-AI is superior to cervical AI, and may have application in Merinos if cervical penetration rates can be improved.