Background: Riociguat and balloon pulmonary angioplasty (BPA) are treatment options for inoperable chronic thromboembolic pulmonary hypertension (CTEPH). However, randomised controlled trials comparing these treatments are lacking. We aimed to evaluate the efficacy and safety of BPA versus riociguat in patients with inoperable CTEPH. Methods: In this phase 3, multicentre, open-label, parallel-group, randomised controlled trial done in 23 French centres of expertise for pulmonary hypertension, we enrolled treatment-naive patients aged 18–80 years with newly diagnosed, inoperable CTEPH and pulmonary vascular resistance of more than 320 dyn·s/cm5. Patients were randomly assigned (1:1) to BPA or riociguat via a web-based randomisation system, with block randomisation (block sizes of two or four patients) without stratification. The primary endpoint was change in pulmonary vascular resistance at week 26, expressed as percentage of baseline pulmonary vascular resistance in the intention-to-treat population. Safety analyses were done in all patients who received at least one dose of riociguat or had at least one BPA session. Patients who completed the RACE trial continued into an ancillary 26-week follow-up during which symptomatic patients with pulmonary vascular resistance of more than 320 dyn·s/cm5 benefited from add-on riociguat after BPA or add-on BPA after riociguat. This trial is registered at ClinicalTrials.gov, NCT02634203, and is completed. Findings: Between Jan 19, 2016, and Jan 18, 2019, 105 patients were randomly assigned to riociguat (n=53) or BPA (n=52). At week 26, the geometric mean pulmonary vascular resistance decreased to 39·9% (95% CI 36·2–44·0) of baseline pulmonary vascular resistance in the BPA group and 66·7% (60·5–73·5) of baseline pulmonary vascular resistance in the riociguat group (ratio of geometric means 0·60, 95% CI 0·52–0·69; p<0·0001). Treatment-related serious adverse events occurred in 22 (42%) of 52 patients in the BPA group and five (9%) of 53 patients in the riociguat group. The most frequent treatment-related serious adverse events were lung injury (18 [35%] of 52 patients) in the BPA group and severe hypotension with syncope (two [4%] of 53 patients) in the riociguat group. There were no treatment-related deaths. At week 52, a similar reduction in pulmonary vascular resistance was observed in patients treated with first-line riociguat or first-line BPA (ratio of geometric means 0·91, 95% CI 0·79–1·04). The incidence of BPA-related serious adverse events was lower in patients who were pretreated with riociguat (five [14%] of 36 patients vs 22 [42%] of 52 patients). Interpretation: At week 26, pulmonary vascular resistance reduction was more pronounced with BPA than with riociguat, but treatment-related serious adverse events were more common with BPA. The finding of fewer BPA-related serious adverse events among patients who were pretreated with riociguat in the follow-up study compared with those who received BPA as first-line treatment points to the potential benefits of a multimodality approach to treatment in patients with inoperable CTEPH. Further studies are needed to explore the effects of sequential treatment combining one or two medications and BPA in patients with inoperable CTEPH. Funding: Programme Hospitalier de Recherche Clinique of the French Ministry of Health and Bayer HealthCare. Translation: For the French translation of the abstract see Supplementary Materials section.