Mutant K-ras provides an independent negative predictive marker for epidermal growth factor receptor (EGFR)-targeted therapy in colorectal cancers (CRCs). Rapid, sensitive, and cost-effective screening for K-ras status will overarch rational personalized medicine. Stool-based DNA testing offers unique advantages for CRC screening such as noninvasiveness, high specificity, and patient compliance, whereas complicated procedures and the low sensitivity of the present approaches have hampered its application on a wide scale. In this study, a chip-based temperature gradient capillary electrophoresis (TGCE) technique was applied to detect low-abundance K-ras mutations under a pooled experiment and analyze K-ras mutations in 30 paired stool samples and cancer tissues of CRC patients and 15 stool samples of healthy volunteers. The chip-based TGCE results showed that the successful analysis of K-ras status could be achieved within 6 min with an extremely low sample consumption of 14 nl. Detection is sensitive enough to reliably report 0.2% mutant CRC cells in a wild-type background, and 0.5 ng of template DNA was sufficient for chip-based TGCE. Of the 30 stool samples of CRC patients analyzed, 17 (57%) harbored K-ras mutations, and the lowest percentage of the detectable mutant K-ras in stool samples was 2%. The coincidence rate for K-ras mutations between stools and tissues obtained by the chip-based method reached 97% (29/30). One of the 15 stool samples of normal controls carried K-ras mutations, producing a specificity of 93%. Clone sequencing data entirely confirmed the results obtained by chip-based TGCE. The study demonstrates that chip-based TGCE is capable of rapidly screening low-abundance K-ras mutations with high sensitivity, reproducibility, simplicity, and significant savings of time and sample. Application of this method to genotype the K-ras gene in stools would provide a potential means for predicting the effectiveness of EGFR-targeted therapy in CRC patients using noninvasive approaches.