In humans, affective states can bias responses to ambiguous information: a phenomenon termed judgment bias (JB). Judgment biases have great potential for assessing affective states in animals, in both animal welfare and biomedical research. New animal JB tasks require construct validation, but for laboratory mice (Mus musculus), the most common research vertebrate, a valid JB task has proved elusive. Here (Experiment 1), we demonstrate construct validity for a novel mouse JB test: an olfactory Go/Go task in which subjects dig for high- or low-value food rewards. In C57BL/6 and Balb/c mice faced with ambiguous cues, latencies to dig were sensitive to high/low welfare housing, environmentally-enriched animals responding with relative ‘optimism’ through shorter latencies. Illustrating the versatility of a validated JB task across fields of research, it further allowed us to test hypotheses about the mood-altering effects of cancer (Experiment 2). Male nude mice bearing subcutaneous lung adenocarcinomas responded more pessimistically than healthy controls to ambiguous cues. Similar effects were not seen in females, however. To our knowledge, this is the first validation of a mouse JB task and the first demonstration of pessimism in tumor-bearing animals. This task, especially if refined to improve its sensitivity, thus has great potential for investigating mouse welfare, the links between affective state and disease, depression-like states in animals, and hypotheses regarding the neurobiological mechanisms that underlie affect-mediated biases in judgment.