KRAS status assessment is mandatory in patients with metastatic colorectal cancer before therapy with anti-epidermal growth factor receptor monoclonal antibodies, as KRAS mutations are associated with resistance to this treatment. However, KRAS genotyping may be very challenging in case of poor tumor cellularity, particularly when major tumor regression is achieved in locally advanced rectal adenocarcinomas after radiochemotherapy. We aimed at identifying the most reliable strategy to detect KRAS mutations in such samples. DNA was extracted from 31 surgical specimens with major tumor regression, following manual dissection, and from paired pre-treatment biopsies and analyzed by high-resolution melting. DNA samples displaying altered melting curve shapes were then sequenced. Samples with unmodified melting curves or wild-type sequence were further investigated by using an allele-specific PCR assay (TheraScreen) and laser microdissection (followed by high-resolution melting and sequencing analyses). In the 31 post-radiochemotherapy surgical specimens, seven KRAS mutations were identified by high-resolution melting analysis/sequencing. One additional mutation was detected by the TheraScreen assay and two mutations, including the one identified by the TheraScreen assay, were detected following laser microdissection. Altogether, 9/31 surgical specimens (29%) presented KRAS mutations. In the manually dissected pre-treatment biopsies, 12 mutations (39%) were identified by high-resolution melting analysis and sequencing. No additional mutations were found by using the TheraScreen assay or laser microdissection. These results indicate that, in the case of post-radiochemotherapy surgical specimens of colorectal cancer with low tumor cellularity, pre-treatment biopsies might represent the most cost-effective option for reliable KRAS genotyping. The use of more sensitive assays, such as allele-specific PCR or laser microdissection, can be envisaged but with higher costs and longer delays.