The objective of this study was to compare the efficiency of artificial insemination (AI) carried out with frozen and fresh, diluted and chilled semen under field conditions. One hundred and twenty-nine mares of different breeds were included in the study. Eighty-one out of the 107 mares inseminated with fresh, chilled semen got pregnant. Seven pregnant mares aborted and 74 foals were born. Out of the 22 mares inseminated with frozen semen, 17 mares got pregnant. Two mares out of the 17 pregnant mares aborted and finally 15 healthy foals were born. No difference was found between the two groups in the ratio of the foals born (P > 0.05). The comparison of medians for the number of insemination cycles did not show significant differences. However, a significant difference (Kruskal-Wallis test, P = 0.014) was found in the number of the inseminations per conception in favour of frozen semen (2.5 vs. 1.8 with fresh chilled and frozen semen, respectively). The Cox regression revealed that the type of semen has a significant impact (P < 0.001) on the service period (duration of the insemination period): the use of frozen semen prolonged the insemination period. This could be due to management issues, since re-insemination with frozen semen took place after only one/a few missed oestrous cycles not used for AI.