To compare the importance of the route of insemination when using fresh or frozen semen, six groups of five bitches were inseminated either into the uterus (groups 4, 5 and 6) or the vagina (groups 1, 2 and 3) with fresh (groups 1 and 4) or frozen semen (groups 2, 3, 5 and 6). The fresh semen was collected when needed from the same dog. The frozen semen used in groups 2 and 5 was obtained from seven dogs on the same day, and pooled and processed simultaneously so that the groups were inseminated with exactly the same semen. The frozen semen used in groups 3 and 6 was obtained from different dogs and processed independently to evaluate not only the effect of the route of insemination but also the potential effect of the dog. The mean concentration of the fresh semen was 310 x 10(6) spermatozoa/ml, its motility was greater than 80 per cent and the percentage of normal live spermatozoa was 80 to 92 per cent. The mean spermatozoal concentration of the frozen semen was 200 x 10(6) spermatozoa/ml, its motility was greater than 60 per cent and the percentage of normal live spermatozoa was 80 per cent. In all the groups there were fewer than 15 per cent abnormal spermatozoa. The animals inseminated with fresh semen received significantly more spermatozoa than the others. The bitches were inseminated twice, three and five days after the estimated peak of luteinising hormone, with a total volume of 5 ml for the vaginal inseminations and 2 ml for the intrauterine inseminations. Sixty per cent of the bitches inseminated with frozen semen and 100 per cent of the bitches inseminated with fresh semen became pregnant, irrespective of the insemination technique used.