BACKGROUND: Around 57 000 people are diagnosed with breast cancer annually in the UK. Radiotherapy can improve local control and survival, but it can also cause long-term side-effects. UK hypofractionation trials have defined dose constraints for heart and lung, reducing heart exposure and treatment burden in a research setting. However, real-world data on heart and lung doses remain scarce. In 2022, the National Health Service (NHS) in England procured a cloud-based radiation dosimetry repository (ProKnow). Using this resource, we aimed to characterise heart and lung doses in routine breast radiotherapy, quantify variation between centres, inform national guidance, and support quality improvement. METHODS: On Dec 11, 2024, NHS England invited all 49 NHS radiotherapy centres in England to upload anonymised radiotherapy plans to ProKnow. Plan upload was stratified by laterality; breast, chest wall, or partial-breast only (26 Gy in five fractions); breast or chest wall and axilla (40 Gy in 15 fractions); and breast or chest wall and axilla and internal mammary chain (40 Gy in 15 fractions). Use of field-based versus volumetric-modulated arc therapy was recorded. Dose-volume heart and lung metrics were analysed against national scorecards derived from UK trials. FINDINGS: Between Dec 11, 2024, and April 24, 2025, 48 of 49 centres participated, contributing 30 582 plans. 27 368 were harvested for analysis, of which 26 236 (95·9%) were analysable. For breast, chest wall, or partial-breast-only treatments, 10 012 (99·8%) of 10 031 left-sided cases and 6961 (99·9%) of 6962 right-sided cases achieved mean heart doses (MHD) below 2 Gy (median MHD 0·57 Gy [IQR 0·43-0·74; left] and 0·25 Gy [0·18-0·36; right]). For nodal plans including internal mammary chain, median MHDs were 3·3 Gy (IQR 2·3-4·5; left) and 2·3 Gy (1·1-3·7; right). 2405 (98·3%) of 2446 plans met the 6 Gy constraint. 8697 (92·1%) of 9448 left-sided breast, chest wall, or partial-breast plans met the optimal lung dose constraint (volume of ipsilateral lung receiving >7·8 Gy being ≤15%) and 1155 (96·0%) of 1203 left-sided nodal-internal mammary chain plans met the mandatory lung dose constraint (volume of lung receiving >17 Gy being ≤35%). Volumetric-modulated arc therapy was used in 65 (0·3%) of 21 119 breast-only plans, 276 (10·5%) of 2638 nodal-non-internal mammary chain plans, and 1738 (70·1%) of 2479 internal mammary chain plans. Volumetric-modulated arc therapy delivered superior target volume coverage but higher MHDs compared to field-based techniques. INTERPRETATION: This breast radiotherapy dosimetry audit shows the feasibility of high-quality large-scale, real-world dosimetric collection and evaluation across the NHS in England. Heart and lung doses predominantly met constraints and were comparable to international benchmarks. However, dosimetric variation existed between centres, particularly in volumetric-modulated arc therapy planning, identifying opportunities for quality improvement. FUNDING: NHS England.