Abstract Background It is mandatory to confirm the absence of mutations in the KRAS gene before treating metastatic colorectal cancers with epidermal growth factor receptor inhibitors, and similar regulations are being considered for non-small cell lung carcinomas (NSCLC) and other tumor types. Routine diagnosis of KRAS mutations in NSCLC is challenging because of compromised quantity and quality of biological material. Although there are several methods available for detecting mutations in KRAS, there is little comparative data regarding their analytical performance, economic merits, and workflow parameters. Methods We compared the specificity, sensitivity, cost, and working time of five methods using 131 frozen NSCLC tissue samples. We extracted genomic DNA from the samples and compared the performance of Sanger cycle sequencing, Pyrosequencing, High-resolution melting analysis (HRM), and the Conformité Européenne (CE)-marked TheraScreen DxS and K-ras StripAssay kits. Results and conclusions Our results demonstrate that TheraScreen DxS and the StripAssay, in that order, were most effective at diagnosing mutations in KRAS. However, there were still unsatisfactory disagreements between them for 6.1% of all samples tested. Despite this, our findings are likely to assist molecular biologists in making rational decisions when selecting a reliable, efficient, and cost-effective method for detecting KRAS mutations in heterogeneous clinical tumor samples.