Nearly two-thirds of military suicides involve firearms, and safe firearm storage is rare. To examine whether US military service members endorse greater openness to safe firearm storage depending on the content of the visual message they are randomly assigned to view. This comparative effectiveness study used a 3 × 2 × 2 factorial design to randomize US military service members to view 1 of 12 visual messages on safe firearm storage. Willingness to use safe firearm storage practices was assessed immediately before and after exposure to the message. Participants were recruited using the KnowledgePanel Calibration approach. Inclusion criteria included current membership in the US military and current firearm ownership. The KnowledgePanel sample was fielded from December 3 to 27, 2021, with a 76% completion rate and 45 individuals determined to be qualified (28% qualification rate). The opt-in sample was fielded December 7, 2021, through January 4, 2022, with 699 individuals (3%) qualified and 674 included in the final data set. Messages shared the same image and text on safe firearm storage but varied in messenger occupation (eg, primary care physician, security forces, or combat controller), the presence of text validating the perspective of firearm owners, and the presence of text validating the drive for home protection. Outcomes included changes in willingness to use 4 at-home (unloaded, separate from ammunition, in a locked location, and with a locking device) and 3 away-from-home (with family or friend, at a firearm retailer, or at a law enforcement agency) firearm storage practices. All analyses, including sample descriptives, are based on weighted data. Of the 719 individuals in the data set, 367 (median [range] age, 33.64 [18-86] years; 80.4% male; 71.4% White) who endorsed not currently storing firearms using the methods assessed were included in analyses. In a multivariate analysis of variance, a significant interaction was found among time, messenger profession, gun-friendly text, and home protection text across all outcomes (Wilks' λ F = 2.09; P = .01; pη2 = 0.040); however, in a post hoc repeated-measures analysis of variance, the interaction was statistically significant only for storing firearms away from home with a trusted family member or friend (F = 5.42; P = .005; pη2 = 0.030). The profession of the messenger was more consistently associated with shifts in willingness than was the message content, although this varied across storage options. The findings of this comparative effectiveness study suggest that several combinations of messenger and content may be associated with willingness to endorse safe firearm storing practices, with particularly consistent positive findings for messages featuring security forces. The scalability and dosage potential of this intervention may render visual messaging valuable for promoting safe firearm storage at the population level.