Objective: We developed a self-report measure to assess men's evaluative attitudes toward violence (i.e., extent to which violence is viewed negatively or positively) and explored the construct validity of its scores. Method: In Study 1, we created a large pool of items, which we administered to two separate online samples of men from the community (N = 318; N = 226). In Study 2, we conducted a randomized experiment as an initial test of the construct validity of the retained items with a new online sample of 510 men from the community. If the measure assesses evaluative attitudes toward violence, then its scores should change in response to established evaluative-attitude-change procedures. Participants were randomly assigned to a condition to make evaluative attitudes toward violence more negative or to a control condition. Results: In Study 1, we retained 17 items to create the Evaluation of Violence Questionnaire (EVQ); selection was based primarily on item variance (to avoid floor effects), interitem correlations (to eliminate redundancy), and factor structure (to ensure unidimensionality). In Study 2, EVQ scores significantly decreased from pre-to posttest for participants in the experimental condition, but did not change for participants in the control condition (comparison of change between conditions: Cohen's d = −0.52, 95% CI [−0.79, −0.25]). Similarly, posttest EVQ scores were significantly lower for participants in the experimental condition than for participants in the control condition (Cohen's d = −0.36, 95% CI [−0.54, −0.17]). Conclusion: These results provide initial support for the construct validity of EVQ scores.