As part of a research program to understand and model eye damage produced by exposure to cw and pulsed lasers, the U.S. Air Force has created a granular model of laser retinal damage. The Thompson granular model simulates absorption of light by melanosomes distributed in the retinal pigmented epithelium, melanosome heating, and subsequent photothermal damage from bulk tissue heating. Various biological input parameters required for the model, such as the density, size, spatial distribution, and absorption coefficient of melanosomes, are not well known, creating uncertainty in the results. This problem is being addressed both experimentally, through measurements of biological parameters for various species, and theoretically, through analysis of parameter sensitivity in the model. In the current study, the parameter sensitivity was analyzed using a technique known as 'design of experiments,' which allows statistical estimation of the relative importance of independent experimental variables. A matrix of 20 cases has been analyzed, using 7 input parameters as independent variables. Cases have been confined to the long pulse regime (greater than or equal to 10 microseconds), where photothermal damage is dominant. Results were assessed using both temperature rise and Arrhenius damage integral values. Corneal fluence was found to be the most important physical parameter and melanosome absorption the most important biological parameter.