We sketch the development of the effective dose approach which provides a theoretical framework to interpret chemotherapy outcome data. Building on the generalised Skipper model of chemotherapy, a meta-regression method is derived to jointly analyse all chemotherapy comparing randomised clinical trials in a given malignancy in order to explore the slope of the effective dose/outcome relationship and the relative potency of cytostatic drugs. The model is applied to explain why treatment differences in aggressive NHL appear to differ by risk groups in aggressive NHL. A respective meta-subgroup analysis to confirm this interaction hypothesis is proposed.