The aim of this study was to find the cut-off points on the visual analogue scale (VAS) to distinguish among mild, moderate, and severe pain, in relation to the following: pain-related interference with functioning; verbal description of the VAS scores; and latent class analysis for patients with chronic musculoskeletal pain. A total of 456 patients were included. Pain was assessed using the VAS and verbal rating scale; functioning was assessed using the domains of the Short Form (36) Health Survey (SF-36). Eight cut-off point schemes were tested using multivariate analysis of variance (MANOVA), ordinal logistic regression, and latent class analysis. The study results showed that VAS scores ⩽ 3.4 corresponded to mild interference with functioning, whereas 3.5 to 6.4 implied moderate interference, and ⩾ 6.5 implied severe interference. VAS scores ⩽ 3.4 were best described for patients with chronic musculoskeletal pain as mild pain, 3.5 to 7.4 as moderate pain, and ⩾ 7.5 as severe pain. Latent class analysis found that a 3-class solution fitted best, resulting in the classes 0.1 to 3.8, 3.9 to 5.7, and 5.8 to 10 cm. Findings from our study agree with those of some other studies, although many other studies found different optimal cut-off point schemes. As there appear to be no universally accepted cut-off points, and in view of the low-to-moderate associations between VAS scores and functioning and between VAS and verbal rating scale scores, the correct classification of VAS scores as mild, moderate. or severe in clinical practice seems doubtful.