Epidemiological factors of tick-borne equine piroplasmoses, caused by Theileria equi and Babesia caballi, were investigated using logistic regression (GLM) and general additive models (GAM) based on the prevalences determined in 510 domestic horses and in ticks in S.W. Mongolia by indirect immunofluorescence antibody test (IFAT) and/or multiplex PCR. Prevalences of T. equi and B. caballi in horses were 66·5% (95% CI: 62·1–70·7) and 19·1% (15·6–22·9), respectively by PCR and 78·8% (74·9–82·3) and 65·7% (61·3–69·9) by IFAT. Of 166 ticks analysed from PCR- and IFAT-negative horses 1 was PCR positive for B. caballi and none for T. equi. GAM demonstrated non-linear increasing proportions of T. equi-PCR and -IFAT positive horses with age suggesting persistent infection. In contrast, the B. caballi-PCR prevalence decreased with age despite a concurrent increase in the proportion of IFAT-positive animals suggesting parasite elimination. The tick (Dermacentor nuttalli) burden of the horses increased with age and decreased with advancing season. Geldings were more likely to be infected with, and seroconvert to, T. equi. Neither herd affiliation, date of sample collection nor abundance of tick infestation had a significant influence on parasite prevalence.