Specific gravity (SpGr) is often used to measure the processing quality of potato (Solanum tuberosum L.) tubers for French fries or potato chips because of its strong correlation with dry matter content and ease of measurement. For French fry processing genotypes, the desirable range for mean SpGr is typically 1.080 to 1.095, and a small variance around the mean is essential for product uniformity. Two multi-year, multilocation trials were conducted to investigate the genetics of SpGr in elite russet germplasm. Consistent with earlier studies, the mean SpGr was measured with high repeatability within each environment: the median plot-basis value was 0.83 for a national trial with six locations and 3 yr. In contrast, the median repeatability of the SD between tubers was only 0.21. Thus, multi-environment trials are needed to identify genotypes with a narrow SpGr distribution. Finlay–Wilkinson stability analysis of the mean SpGr established one genotype as an outlier: when best linear unbiased predictions were regressed on the environment means, this genotype had a regression coefficient of 2.1, compared with 0.4 to 1.4 for the others. The genetic correlation between environments showed a consistent regional pattern in mean SpGr over the years. There was a higher mean correlation between environments within the Pacific Northwest (0.97), Upper Midwest (0.91), and Northeast (0.85) than between environments from the different regions (0.35–0.78). Although breeding for national adaptation is an attractive idea, our results suggest that genetic gain may be easier to achieve at the regional level.