Early pregnancy loss (EPL) between days 15-65 after breeding has been shown to occur in 7.9% of equine pregnancies with substantial economical, welfare and safety implications. Whilst maternal age has been recognised as an important risk factor in relation to the incidence of EPL, few other risk factors have been conclusively identified. Further, multivariable data analysis of risk factors for EPL is sparse. A prospective cohort investigation of thoroughbred broodmares in the United Kingdom was conducted over the 2013 and 2014 breeding seasons. Information relating to 28 factors including mare, stallion, pregnancy and therapeutic interventions was collected using questionnaires and entered into a custom-designed Microsoft Access database. Mixed effects logistic regression was used to determine risk factors for EPL, including 'mare' as a random effect to account for repeat pregnancies in the same mare. Stallion, stud and veterinarian were also evaluated as random effects. Variables with a p-value of <0.25 in univariable analysis were taken forward for consideration in the multivariable model which was built using a forward stepwise approach. Data were collected on 2245 pregnancies in 1753 mares. Increasing mare age (OR = 1.11, 95% confidence interval (CI) = 1.04, 1.18, p = 0.001), having had one previous foal (OR = 3.52, 95% CI = 1.56, 7.95, p = 0.002) and presence of uterine cysts (OR = 1.76, 95% CI = 1.07, 2.91, p = 0.03) were all associated with increased odds of EPL following multivariable analysis. Increasing day 15/16 scan vesicle size (OR = 0.24, 95% CI = 0.16, 0.38, p < 0.001) and the use of ovulatory induction agents (OR = 0.31, 95% CI = 0.17, 0.55, p < 0.001) were negatively associated with EPL. Stallion, stud and veterinarian were not significantly associated with EPL. Analysis of a subpopulation of 344 multiple (twin and triplet) pregnancies found that the use of flunixin meglumine at the time of manual reduction of a multiple pregnancy resulted in reduced odds of EPL (OR = 0.34, 95% CI = 0.14, 0.84, p = 0.02). Results from this study can be used by stud farm personnel when assessing their broodmare population and by clinicians when deciding upon therapeutic strategies. Additional work can be focused around these risk factors to further our understanding of the pathophysiology of EPL.