The purpose of this pilot study was to explore the feasibility of using hand drawn images to identify symbol components for incorporation into warning symbol design software. This software will use an interactive evolutionary computation (IEC) algorithm to generate and evolve symbols mathematically described by a set of numerical parameters. Therefore, participants (N = 100) ages 19–43 (x = 23.2) were recruited to determine these symbol design parameters. Participants were invited to hand draw warning symbols for three referents: fall from elevation, hearing protection, and hazardous atmosphere. A panel of design engineers determined 27 attributes were present in the fall from elevation, 19 in the hearing protection, and 25 in the hazardous atmosphere images. A direct clustering algorithm was used to determine which attributes, or symbol parameters, were most commonly present or conspicuously absent among the clustered image families. For the fall from elevation, hearing protection and hazardous atmosphere referents, the clustering algorithm identified six, four and four symbol parameters, respectively, primarily responsible for distinguishing one drawn symbol from another. Thus, these parameters will be included as evolvable genes in the IEC software.