A pregnancy of unknown location (PUL) describes a clinical situation where a woman presents with a positive urine pregnancy test but the location of the pregnancy cannot be determined when a transvaginal ultrasound scan is performed. The concern is that the patient may be harbouring an ectopic pregnancy (a pregnancy located outside of the uterus, most commonly in the fallopian tube), which remains the leading cause of early-pregnancy related maternal mortality in the UK. There remains a great deal of heterogeneity in the way women with a PUL are managed and this is not always evidence-based. There is therefore an unmet clinical need to improve the care of these patients. This manuscript outlines original research on how biomarkers, both novel and those in current clinical use, can be best used to optimise the clinical care women with a PUL receive. Participants classified as a PUL at their initial ultrasound scan were recruited and underwent serial ultrasound scans and serum blood sampling in the first trimester. Women underwent routine follow-up and treatment as was clinically indicated. Those participants recruited to study 5 (see below) underwent additional bio-fluid sampling (urine and vaginal swabs, as well as serum blood testing). Pregnancy outcomes were collected using hospital paper records and IT systems. 7,139 women were recruited to the studies in total. Five separate analyses were conducted to meet the aims of this thesis: Study 1: I performed a systematic review and meta-analysis to review the current evidence base for managing women with a pregnancy of unknown location. A total of 8,847 titles were screened and 43 studies analysed, with 23,802 women with a PUL recruited by these studies in total. This analysis demonstrated a logistic regression model called the M4 model was the current best diagnostic protocol for managing women with a PUL and predicting a final outcome of ectopic pregnancy. Study 2: I undertook a clinical implementation study involving 1,022 patients initially classified as a PUL across three centres in the UK to assess the test performance and safety profile of using the M4 model in ‘real-time’ clinical practice. It demonstrated good test performance (classifying 70% PUL as low-risk after just two hospital visits, with a negative predictive value of 97% and a sensitivity for correctly predicting ectopic pregnancy at 82%). Study 3: As there remained scope for improvement in test performance, a newer logistic regression model (the M6 model) was developed on 2,753 patients from two UK centres. We carried out a diagnostic accuracy study which demonstrated the M6 model had superior test performance to the M4 model, classifying 62% PUL as low-risk after just two hospital visits, with a negative predictive value of 99% and a sensitivity for correctly predicting ectopic pregnancy at 92%. Study 4: I undertook a clinical implementation study involving 3,272 patients initially classified as a PUL across eight centres in the UK to assess the test performance and safety profile of using the M6 model (as part of a two-step management protocol) in ‘real-time’ clinical practice. It demonstrated good test performance (classifying 16% PUL as low-risk after just one hospital visit and a further 54% after just two visits with a negative predictive value of 98% and 99% respectively and a sensitivity for correctly predicting ectopic pregnancy at 86%). There were no clinical safety issues. Study 5: I analysed the vaginal microbial composition in 92 women initially classified as a PUL. This demonstrated reduced Lactobacillus species abundance in pregnancies with a final outcome of an ectopic pregnancy compared to controls (i.e. those with a viable intrauterine pregnancy) that was independent of vaginal bleeding and preceded diagnosis of the final pregnancy outcome. Summary conclusions: The studies carried out as part of this thesis demonstrate that a two-step triage protocol consisting of: Step 1 – a progesterone cut-off of 2nmol/ l and Step 2 – the M6 model is the most effective way to manage women with a PUL and accurately predict a final outcome of an ectopic pregnancy. These findings can be used to rationalize the management of these women so they can be given more consistent and evidence-based care. Novel biomarker studies looking at the vaginal microbiota have shown promising preliminary results but further work is required in this area to establish the mechanistic pathways underlying the pathophysiology of ectopic pregnancy.