The aim of the present study was to test the hypothesis that high serum YKL-40 associates with colorectal cancer in subjects at risk of colorectal cancer. We measured serum YKL-40 in a prospective study of 4,496 Danish subjects [2,064 men, 2,432 women, median age 61 years (range, 18-97)] referred to endoscopy due to symptoms or other risk factors for colorectal cancer. Blood samples were collected just before large bowel endoscopy. Serum YKL-40 was determined by ELISA. Serum YKL-40 was higher (P < 0.0001, unadjusted for confounding covariates) in subjects diagnosed with colon cancer (median 126 μg/L, 25%-75%: 80-206 μg/L) and rectal cancer (104, 72-204 μg/L) compared with subjects with adenoma (84, 53-154 μg/L), other nonmalignant findings (79, 49-138 μg/L), and no findings (62, 41-109 μg/L). Serum YKL-40 independently predicted colorectal cancer [OR, 1.53; 95% confidence interval (CI), 1.40-1.67; AUC = 0.68, P < 0.0001]. Restricting the analysis to subjects with no comorbidity increased the OR for serum YKL-40 to predict colorectal cancer (OR, 1.82; 1.58-2.08; AUC = 0.73, P < 0.0001). Combining serum YKL-40 and CEA demonstrated that both were significant [(YKL-40, OR, 1.27; 95% CI, 1.16-1.40); (CEA, OR, 1.92; 1.75-2.10; AUC = 0.75, P < 0.0001; OR for a 2-fold difference in marker level)]. Multivariable analysis (YKL-40, CEA, age, gender, body mass index, and center) showed that serum YKL-40 was a predictor for colorectal cancer in individuals without comorbidity (OR, 1.25; 95% CI, 1.05-1.40; P = 0.012), whereas this was not the case for those with comorbidity (OR, 0.98; 95% CI, 0.84-1.14; P = 0.80). In conclusion, high serum YKL-40 in subjects suspected of colorectal cancer and without comorbidity associates with colorectal cancer. Determination of serum YKL-40 may be useful in combination with other biomarkers in risk assessment for colorectal cancer. Cancer Epidemiol Biomarkers Prev; 24(3); 621-6. ©2015 AACR.