This retrospective study utilized a mixed-effects logistic model analysis to investigate variables associated with the probability of pregnancy loss (PL) between days 30 and 70 in a dataset comprising 9507 pregnancies from a single cattle herd over 10 years. The model incorporated fixed-effect variables including cow breed (Holstein, Crossbred, and Brown Swiss), parity (1st, 2nd, 3rd, and 4th or more), insemination seasons, insemination number (≤3 vs >3), estrus nature (spontaneous vs synchronized), postpartum problems, sire breed (Holstein, Fleckvieh, Brown Swiss, and Montbéliarde), zygote genotypic (pure vs crossbred), days in milk (DIM) at insemination, actual 305-day milk yield, and sire conception rate. Additionally, random effects included sire (n = 129), cow (n = 3463), and production years (n = 10). The results revealed that cows inseminated with Brown Swiss sires or sires with lower initial conception rates had higher PL rates. Biparous cows, cows with lower 305-day milk yield, cows inseminated later in DIM, cows receiving the 4th or subsequent insemination, and cows inseminated during winter or autumn had lower PL rates. The estrus type and zygote genotype did not significantly impact PL. The random effects of cow, sire, and production years were estimated at 0.230, 0.054, and 0.112, respectively. In conclusion, the study findings suggested that improving management practices for high-yielding cows, cows in early lactation stages, and cows exposed to thermal stress conditions, along with utilizing Brown Swiss cows and sires with high initial conception rates per insemination, could potentially decrease overall PL rates on the farm. Nonetheless, the results did not support the use of Crossbred cows, sires from different breeds, or specific sires to mitigate PL rates on the farm.