There has been much research on how people use instructional information to gain procedural knowledge. In the context of procedures involving physical objects, however, there has been little research on the role these objects play in conveying procedural information. This study investigated how people used instructions – presented as either images or text – to assemble various physical objects. Objects were selected that either comprised uniquely shaped or interchangeable parts. Participants assembled each object twice, randomly receiving either image or text instructions for each build. They then assembled each object without the instructions and made judgments about the order of the procedure from memory. Image instructions generally resulted in faster and more accurate assemblies as well as more accurate memory for procedural order. These results were found only for objects with uniquely shaped parts. An object comprising interchangeable parts was readily assembled with either instructional type. Although text alone failed to provide any advantages, the combination of images and then text resulted in more consistent mental workload, which might be beneficial in some operational contexts. These results provide insights about how physical objects influence the use of and knowledge gained from procedural instructions.