It is important to estimate the biological characteristics of tumors, including the nodal status at the time of diagnosis for optimal treatment of individual cancer patients. Array-based comparative genomic hybridization (aCGH) was performed on 77 sporadic colorectal adenocarcinomas using a chip spotted with 4030 BAC clones. The nodal status was compared with an aCGH profiles depicted using a combination of decision-tree classifier and a Self-Organizing Map (SOM) analysis. Node metastasis was not detected in any of the 6 poorly differentiated adenocarcinomas with a 3q loss. A SOM analysis following the decision-tree classification of the aCGH data allowed for the differentiation in chromosomal regions between high- and low-level decreases in the DNA copy number. Node metastasis was detected in all 5 tumors with the high-level decrease in DNA copy number at Xp, irrespective of the histological type. Node metastasis was also found exclusively in 6 tumors with increase in DNA copy number at the chromosomal region between 11q13.3 and 11q22.3. Copy number aberrations linked to nodal metastasis were identified more collectively by the combination of the decision-tree classifier and a SOM analysis than by the conventional analysis method in aCGH analysis.