In four Kenyan pig breeding units the pregnancy diagnosis of sows has been carried out in two groups: Group 1 (n = 1911): the sows were transrectaly pregnancy tested between Days 17–22 post-mating by ultrasound. Sows testing non-pregnant immediately received one dose of 400 IU pregnant mare serum gonadotropin (PMSG) (equine chorion gonadotropin, eCG) and 200 IU human chorion gonadotropin (hCG). On showing signs of oestrous, the animals were subsequently artificially inseminated (AI). Group 2 (n = 1923): sows were pregnancy tested by serum progesterone (P4)-based enzyme-linked immunosorbent assay (ELISA) on Day 17 post-breeding. P4 concentrations were categorized as positive (> 5 ng/ml) or negative (< 5 ng/ml). Sows testing non-pregnant immediately received one dose of 400 IU PMSG and 200 IU hCG by injection, and were subsequently artificially inseminated. The following parameters were evaluated: sows diagnosed non-pregnant, days from first post-weaning insemination until the sows were inseminated at their first return to oestrus; farrowing rate and total piglets born and number of live-born piglets in litters. The percentage of sows diagnosed non-pregnant in the two groups, as well as the totals of born piglets and of live-born piglets in litters did not differ significantly between the two groups. The number of days from the first post-weaning mating until the sows were artificially inseminated at their first return to oestrus and the administration of eCG and hCG was shorter (P < 0.01) and farrowing rate was higher (P < 0.01) in the ELISA-tested sows.