1) Background: Ultrasound-based strain imaging is now available in several ultrasound (US) scanners. Strain ratio (SR) can be used to quantify strain recorded simultaneously in two different user-selected areas, ideally exposed to the same amount of stress. The aim of this study was to evaluate SR variability when assessed in an in-vitro setup with a tissue-mimicking phantom, on resected tissue samples and in live tissue scanning with endoscopic applications. 2) Material and methods: Retrospective analysis of SR for quantification of elastic contrasts in a tissue-mimicking phantom containing four homogenous inclusions, in 38 resected bowel wall lesions and in 48 focal pancreatic lesions examined in vivo. Median SR and the inter-quartile range (IQR) was calculated on all external and endoscopic US-applications. The IQR/median provides a measure of the SR variability focusing on the two percentiles of the data closest to the median value. 3) Results: The overall variability of SR was lowest in a tissue-mimicking phantom (mean QR/median SR: 0.07). In resected bowel wall lesions representing adenomas, adenocarcinomas or Crohn lesions, the variability increased (mean IQR/Median: 0.62). During an endoscopic examination of focal pancreatic lesions in vivo, the variability increased further (mean IQR/Median: 2.04). 4) Conclusion: SR variability increased when assessed on different targets with growing heterogeneity and biological variability as one moved from homogeneous media to live tissues and endoscopic application. This may indicate a limitation for the accuracy SR evaluation in clinical applications.