The most critical phase of pregnancy is the first three weeks following insemination. During this period about 50% of high yielding lactating cows suffer embryonic loss prior to implantation, which poses a high economic burden on dairy farmers. Early diagnosis of pregnancy in cattle is therefore essential for monitoring breeding outcome and efficient production intervals. Regulated microRNAs (miRNAs) that reach easily accessible body fluids via a 'liquid biopsy' could be a new class of pregnancy predicting biomarkers. As milk is obtained regularly twice daily and non-invasively from the animal, it represents an ideal sample material. Our aim was to establish a pregnancy test system based on the discovery of small RNA biomarkers derived from the bovine milk cellular fraction and skim milk of cows. Milk samples were taken on days 4, 12 and 18 of cyclic cows and after artificial insemination, respectively, of the same animals (n = 6). miRNAs were analysed using small RNA sequencing (small RNA Seq). The miRNA profiles of milk cells and skim milk displayed similar profiles despite the presence of immune cell related miRNAs in milk cells. Trends in regulation of miRNAs between the oestrous cycle and pregnancy were found in miR-cluster 25~106b and its paralog cluster 17~92, miR-125 family, miR-200 family, miR-29 family, miR-15a, miR-21, miR-26b, miR-100, miR-140, 193a-5p, miR-221, miR-223, miR-320a, miR-652, miR-2898 and let-7i. A separation of cyclic and pregnant animals was achieved in a principal component analysis. Bta-miRs-29b, -221, -125b and -200b were successfully technically validated using quantitative real-time PCR, however biological validation failed. Therefore we cannot recommend the diagnostic use of these miRNAs in milk as biomarkers for detection of bovine pregnancy for now.