Heterogeneity of DNA content was analyzed in 389 samples from 65 resected gastric cancers. Analysis of the samples revealed that there were 14 homogeneously diploid tumours. Six tumours were uniformly DNA aneuploid, each tissue block containing the same DNA index. The other 45 tumours (69%) varied in DNA content heterogeneity. In 39 of 45 tumours, there was a mixture of diploid and aneuploid samples, and 25 of the 39 tumours had a single aneuploid stemline. In 14 out of 39 tumours, there was also a mixture of diploid and aneuploid samples having two or more DNA aneuploid stemlines. In the remaining six tumours, different DNA aneuploid stemlines were contained in different samples without evidence of diploidy. When four or fewer samples were analyzed, only 50% of the tumours were diagnosed as having DNA content heterogeneity. On the other hand, 78% of the tumours showed DNA heterogeneity when 5 or more samples were analyzed. If the tumours had not been widely sampled, about a quarter of the tumours would have been mislabeled as diploid. The patients with tumours showing homogeneous diploidy survived longer than those with tumours showing a mixture of diploid and aneuploid stemlines. The survival rate was lowest for the patients with tumours having a mixture of diploid and multiple aneuploid stemlines, compared with those showing homogeneous diploid or a mixture of diploid and single aneuploid stemlines. The data from the current study clearly demonstrate the importance of adequate sampling in assessing the ploidy status of gastric cancers to identify groups of patients running different clinical course and prognosis.