Background: Oncogenic alterations in RET represent important therapeutic targets in thyroid cancer. We aimed to assess the safety and antitumour activity of pralsetinib, a highly potent, selective RET inhibitor, in patients with RET-altered thyroid cancers. Methods: ARROW, a phase 1/2, open-label study done in 13 countries across 71 sites in community and hospital settings, enrolled patients 18 years or older with RET-altered locally advanced or metastatic solid tumours, including RET-mutant medullary thyroid and RET fusion-positive thyroid cancers, and an Eastern Co-operative Oncology Group performance status of 0–2 (later limited to 0–1 in a protocol amendment). Phase 2 primary endpoints assessed for patients who received 400 mg once-daily oral pralsetinib until disease progression, intolerance, withdrawal of consent, or investigator decision, were overall response rate (Response Evaluation Criteria in Solid Tumours version 1.1; masked independent central review) and safety. Tumour response was assessed for patients with RET-mutant medullary thyroid cancer who had received previous cabozantinib or vandetanib, or both, or were ineligible for standard therapy and patients with previously treated RET fusion-positive thyroid cancer; safety was assessed for all patients with RET-altered thyroid cancer. This ongoing study is registered with clinicaltrials.gov, NCT03037385, and enrolment of patients with RET fusion-positive thyroid cancer was ongoing at the time of this interim analysis. Findings: Between Mar 17, 2017, and May 22, 2020, 122 patients with RET-mutant medullary and 20 with RET fusion–positive thyroid cancers were enrolled. Among patients with baseline measurable disease who received pralsetinib by July 11, 2019 (enrolment cutoff for efficacy analysis), overall response rates were 15 (71%) of 21 (95% CI 48–89) in patients with treatment-naive RET-mutant medullary thyroid cancer and 33 (60%) of 55 (95% CI 46–73) in patients who had previously received cabozantinib or vandetanib, or both, and eight (89%) of nine (95% CI 52–100) in patients with RET fusion-positive thyroid cancer (all responses confirmed for each group). Common (≥10%) grade 3 and above treatment-related adverse events among patients with RET-altered thyroid cancer enrolled by May 22, 2020, were hypertension (24 patients [17%] of 142), neutropenia (19 [13%]), lymphopenia (17 [12%]), and anaemia (14 [10%]). Serious treatment-related adverse events were reported in 21 patients (15%), the most frequent (≥2%) of which was pneumonitis (five patients [4%]). Five patients [4%] discontinued owing to treatment-related events. One (1%) patient died owing to a treatment-related adverse event. Interpretation: Pralsetinib is a new, well-tolerated, potent once-daily oral treatment option for patients with RET-altered thyroid cancer. Funding: Blueprint Medicines.