Two retroviral DNAs that encode the normal human epidermal growth factor (EGF) receptor hEGFR have been generated by inserting a hEGFR cDNA into two different retroviral vectors. One DNA (pCO11-EGFR-neo) also contained a linked selectable marker gene (neoR). The other (pCO12-EGFR) only expresses hEGFR. When introduced into NIH3T3 cells, the two DNAs and the viruses derived from them induced a fully transformed phenotype, including focal transformation and growth in agar or low serum, but transformation depended entirely upon EGF being present in the growth medium. Compared with pCO11-EGFR-neo, pCO12-EGFR induced EGF-dependent transformation 2-5 times more efficiently and expressed higher numbers of receptors (4 x 10(5) vs. 1 x 10(5) EGF receptors per cell). The results indicate that transforming potential is directly related to the number of EGF receptors. In defined, serum-free medium that contained only very low concentrations of insulin (0.6 microgram/ml) and transferrin (0.6 micrograms/ml), hEGFR-virus infected cells were able to grow with EGF as the only growth factor. Moreover, daily incubation of the cells with EGF for only 30 min was sufficient to induce growth. NR6 cells, which lack endogenous EGF receptors, were transformed as efficiently as NIH3T3 cells by the hEGFR virus. The dose-dependent growth response to EGF of infected NR6 cells grown in serum-free medium can be used as a highly sensitive bioassay for the quantitative assessment of EGF and transforming growth factor type alpha (TGF alpha). This bioassay is at least as sensitive as previously reported radioimmunoassays and can measure a much wider concentration range (10 pg-100 ng/ml). Uninfected NR6 cells or NR6 cells infected by helper virus alone can be used as controls for the EGF specificity of growth stimulation.