OBJECTIVES: The EQ-5D visual analog scale (VAS) is a patient-reported rating of overall health that is often reported in clinical studies. However, few studies are available to guide in the interpretation of meaningful differences in VAS scores. The objective of this study was to estimate meaningful differences in EQ-5D VAS scores in cancer, particularly in lung cancer. METHODS: Secondary data analysis was conducted on a cross sectional study of 534 cancer patients, including 50 lung cancer patients, who completed EQ-5D VAS (scaled from 0 (worst imaginable health) to 100 (best imaginable health)). Anchor-based and distribution based approaches were used to estimate important differences for VAS scores. Cancer patients were grouped into clinically meaningful categories anchored by: 1) Eastern Cooperative Oncology Group performance status (PS), and 2) FACT-G total score-based quintiles. These anchors were conservative partitions likely to exceed the true minimum important difference (MID). Distribution-based criteria applied to each subgroup included 1/2 standard deviation (SD) and the standard error of the measure (SEM). RESULTS: Estimates of MID for VAS scores based on PS categories ranged from 8 (average mean difference across categories) to 11 (SEM) for all cancer patients, and from 8 (0.5 SD) to 12 (average mean difference across PS categories) for lung cancer patients. Using FACT-G score quintiles, MIDs were the same for both the overall cancer groups and the lung cancer subgroup where the average mean difference between quintiles was 7, SEM was 10 and 1/2 SD was 9. CONCLUSION: The range of estimates representing important differences in EQ-5D VAS scores was similar between all cancers and lung cancer (7 to 12), with the lower bounds of MID estimates closer to minimal important differences, i.e. 7-8. These estimates can help to inform interpretation of EQ-5D VAS scores, particularly in studies of cancer.