Curative cancer therapies are uncommon and nearly always involve multi-drug combinations developed by experimentation in humans; unfortunately, the mechanistic basis for the success of such combinations has rarely been investigated in detail, obscuring lessons learned. Here we use isobologram analysis to score pharmacological interaction, and clone tracing and CRISPR screening to measure cross-resistance among the five drugs comprising R-CHOP, a combination therapy that frequently cures Diffuse Large B-Cell Lymphomas. We find that drugs in R-CHOP exhibit very low cross-resistance but not synergistic interaction; together they achieve a greater fractional kill according to the null hypothesis for both the Loewe dose-additivity model and the Bliss effect-independence model. These data provide direct evidence for the 50-year old hypothesis that a curative cancer therapy can be constructed on the basis of independently effective drugs having non-overlapping mechanisms of resistance, without synergistic interaction, which has immediate significance for the design of new drug combinations. Lymphoma (DLBCL). R-CHOP has five constituents: R -rituximab, a humanized monoclonal antibody against CD20, a protein expressed on the surface of all B cells; C -cyclophosphamide (Cytoxan®) an alkylating agent; H -hydroxydaunomycin (doxorubicin, or Adriamycin®), a topoisomerase II inhibitor; O -Oncovin® (vincristine), an anti-microtubule drug and; P -prednisone, a steroid. R-CHOP was developed over an extended period of time via clinical experiments in humans (Lakhtakia and Burney, 2015). The constituents of R-CHOP are known to be individually cytotoxic to DLBCL cells in vivo, and the drugs have largely non-overlapping dose-limiting toxicities, which permits their combined administration in patients. The reasons for the clinical superiority of R-CHOP in DLBCL remain poorly understood. Pritchard et al. (2013) observed no synergy among pairs of drugs in CVAD (similar to CHOP) in a mouse cell line model of Non-Hodgkin lymphoma, and in profiling the effects of 93 gene knockdowns by RNA interference on drug sensitivity, the change in sensitivity to CVAD was equal (for almost every knockdown) to the average of its changes in single-drug sensitivity; this demonstrates that CVAD does not act as a more potent version of a single drug, nor does it exhibit a new signature of genetic dependencies.We tested for pharmacological interaction among all pairs of R-CHOP constituents across a full dose range in three DLBCL cell lines and assessed interaction using both the Bliss independence and Loewe additivity criteria. We observed little if any synergy: most drug pairs were additive and some were antagonistic. We also tested higher order combinations at fixed dose ratios with similar results. We then screened for cross resistant mutations using random mutagenesis with clone tracing as well as CRISPR interference (CRISPRi) and CRISPR activation (CRISPRa) with genome-scale libraries. The rate of multi-drug resistance was near the theoretical minimum pr...