Simple Summary: Mice are widely used to study various types of cancer to develop effective treatments, but this can cause them to experience pain or poor welfare, which may lead to inconsistent or negative findings. This study used a variety of welfare monitoring methods to determine if, or more importantly when mice developing lung cancer experienced poor welfare so that similar future studies can be ended before this occurs. It also aimed to determine whether non-aversive handling, also known as tunnel handling (lifting mice in a plastic tube rather than by the tail) is an effective method of refinement, leading to improved welfare and improved consistency of results. Compared to normal mice, those developing lung cancer showed behaviour changes consistent with poor welfare including a general lack of movement and facial expression changes (grimacing). Most noticeably, cancer caused a reduction in food consumption and the ability of the mice to build a suitable nest. These combined factors suggested the mice began to suffer around 4 days before the study ended. Tunnel handling did not improve our results consistency or enhance welfare, but importantly, it had no negative effects. The findings suggest similar studies should be ended when mice appear to make reduced quality nests. Cancer-bearing mice are at risk of developing anxiety, pain, or malaise. These conditions may not only harm welfare but could also undermine data quality and translational validity in studies to develop therapeutic interventions. We aimed to establish whether, or at what point mice developing lung cancer show these symptoms, what measures can best detect their onset, and if data quality and animal welfare can be enhanced by using non-aversive handling (NAH). Welfare was monitored using various daily methods. At the beginning and end of the study, we also scored behaviour for general welfare evaluation, recorded nociceptive thresholds, and applied the mouse grimace scale (MGS). Cancer caused a decline in daily welfare parameters (body weight, and food and water consumption) beginning at around 4 days prior to euthanasia. As cancer progressed, rearing and walking declined to a greater extent in cancer-bearing versus control mice, while grooming, inactive periods, and MGS scores increased. A decline in nest building capability and food consumption provided a particularly effective means of detecting deteriorating welfare. These changes suggested a welfare problem arose as cancer developed, so similar studies would benefit from refinement, with mice being removed from the study at least 4 days earlier. However, the problem of highly varied tumour growth made it difficult to determine this time-point accurately. There were no detectable beneficial effects of NAH on either data quality or in terms of enhanced welfare. [ABSTRACT FROM AUTHOR]