Objectives were to determine associations between percentage pregnancy loss (PPL) in dairy cattle and: (i) pregnancy diagnosis by ultrasonography; (ii) pregnancy diagnosis by serum pregnancy-specific protein B (PSPB) concentrations, with or without serum progesterone concentrations; and (iii) production and environmental factors. This study included 149 822 pregnancy diagnoses conducted over 13 years in Holstein-Friesian cows in Hungarian dairy herds. The following were determined: PPL in cows diagnosed pregnant by transrectal ultrasonography 29–42 days after artificial insemination (AI; n = 11 457); PPL in cows diagnosed pregnant by serum PSPB 29–35 days after AI (n = 138 365); and PPL and its association with serum progesterone concentrations, PSPB and production/environmental variables. The definition of PPL was percentage of cows initially diagnosed pregnant based on ultrasonography or PSPB, but not pregnant when examined by transrectal palpation 60 –70 days after AI. The PPL was lower (p < 0.001) in cows following ultrasonographic vs PSPB diagnosis of pregnancy at 29–35 days (8.1 vs 19.3%, respectively), but was higher in cows following ultrasonographic pregnancy diagnosis on 29–35 vs 36–42 days (8.1 vs 7.1%, respectively, P < 0.05). Furthermore, 72.9% of pregnancies with ultrasound-detected morphological abnormalities resulted in pregnancy loss. As a subset of PSPB data, a fully quantitative PSPB assay was used for 20 430 samples; PPL in cows with a high PSPB concentration (>1.1 ng/ml) was lowest (15.0%), whereas cows with low concentrations of both PSPB and progesterone (0.6–1.1 and <2 ng/ml, respectively) had the highest PPL (76.3%; p < 0.0001). Furthermore, PPL was higher in cows with advanced parity and with high milk production, when ambient temperatures were high, although body condition score (BCS) had no effect on PPL. Finally, there were no significant associations between serum PSPB and environmental temperatures or number of post-partum uterine treatments.