The assessment of animal welfare is reliant upon the application of animal- and resource-based indicators. Animal-based indicators (physical, physiological and behavioural) are considered to be more representative of an animal's welfare state but are more difficult for an assessor to interpret. In order to build a robust composite framework for the assessment of welfare of dogs (Canis familiaris) within Catch-Neuter-Return (CNR) programmes, including both resource- and animal-based indicators, it is necessary to first evaluate whether appropriate capture and handling techniques plus behavioural, animal-based indicators can be reliably assessed by staff working in CNR programmes. Results of a video-based survey of experienced dog management staff and CNR practitioners are reported and indicate that staff experienced in canine CNR are reliably able to agree on acceptable and unacceptable handling and capture techniques. However, there is only fair agreement between observers in positively recognising pain. Thus, dog welfare in CNR may be at risk unless staff are effectively trained at recognising behavioural indicators of poor welfare including pain behaviours. This paper suggests that non-invasive, visual indicators of pain, such as facial tension and body posture, may be a reliable and effective approach to recognising post-operative pain in street dogs.