In clinic, cetuximab, an anti-EGFR antibody, improves treatment outcomes in colorectal cancer (CRC). KRAS-mutant CRC is generally resistant to cetuximab, although difference of the sensitivity among KRAS-mutants has not been studied in detail. We previously developed the cancer tissue-originated spheroid (CTOS) method, a primary culture method for cancer cells. We applied CTOS method to investigate whether ex vivo cetuximab sensitivity assays reflect the difference in sensitivity in the xenografts. Firstly, in vivo cetuximab treatment was performed with xenografts derived from 10 CTOS lines (3 KRAS-wildtype and 7 KRAS mutants). All two CTOS lines which exhibited tumor regression were KRAS-wildtype, meanwhile all KRAS-mutant CTOS lines grew more than the initial size: were resistant to cetuximab according to the clinical evaluation criteria, although the sensitivity was quite diverse. We divided KRAS-mutants into two groups; partially responsive group in which cetuximab had a substantial growth inhibitory effect, and resistant group which exhibited no effect. The ex vivo signaling assay with EGF stimulation revealed that the partially responsive group, but not the resistant group, exhibited suppressed ERK phosphorylation ex vivo. Furthermore, two lines from the partially responsive group, but none of the lines in the resistant group, exhibited a combinatory effect of cetuximab and trametinib, a MEK inhibitor, ex vivo and in vivo. Taken together, the results indicate that ex vivo signaling assay reflects the difference in sensitivity in vivo and stratifies KRAS mutant CTOS lines by sensitivity. Therefore, coupling the in vivo and ex vivo assays with CTOS can be a useful platform for understanding the mechanism of diversity in drug sensitivity.