In this overview we review and model how radiotherapy tumour control and complication rates vary with dose, fractionation, schedule duration, irradiated volume and use of chemotherapy for stage III non-small cell lung cancer (NSCLC), and use the modelling to study the effectiveness of different NSCLC dose-escalation approaches being developed in the UK. Data have been collated for pneumonitis, lung fibrosis, early and late oesophagitis, cord and cardiac complications, and local progression-free survival at 30 months. Dependences of the various end points on treatment-related factors are catalogued and analysed using the linear-quadratic incomplete repair model to account for dose and fractionation effects, making linear corrections for differences in schedule duration, and loosely characterising volume effects using parallel- and series-type concepts. Tolerance limits are calculated for the different end points and distilled into ranges of prescribed dose likely to be tolerable when delivered in 2.5 and 4 week radiation and 6 week chemoirradiation schedules using conformal techniques. Worthwhile ( approximately 20%) gains in 30 month local progression-free survival should be achievable at safely deliverable levels of dose escalation. The analysis suggests that longer schedules may be more beneficial than shorter ones, but this finding is governed by the relative rates of tumour and oesophageal accelerated proliferation, which are quite imprecisely known. Consequently escalated 2.5, 4 and 6 week schedules are being developed; each should lead to useful improvements in local control but it is not yet known which schedule will be most effective.