pukaskwa national park has implemented a natural fire regime emulation program. one component of the natural fire regime is the pattern of ignitions, primarily from lightning. there were three ways in which patterns of lightning strikes were investigated: a geographical analysis, a fire weather analysis, and an ignition analysis. the analysis for the first two components was carried out using a kernel analysis, which is a non-parametric probability density estimation technique. in the geographic analysis, probability of lightning strikes was tabulated by characteristics of the landscape- vegetation, topography, ecodistrict, and proximity to lake superior. it was shown that there was a strong effect of the proximity to lake superior- areas near the lake have a higher probability of lightning strikes than areas inland. the topographic component showed that high points on the landscape were more likely to get struck than the rest of the landscape. other effects were masked by the effect of proximity to lake superior. in the fire weather analysis, probability density surfaces were prepared from the daily weather and from the weather on the day of lightning strikes. strikes are most frequent at values of the fine fuel moisture code (ffmc) and duff moisture code (dmc) that are notconducive to the ignition and survival of fire. the exception to this was a storm in june, 1995 which happened at a high value of ffmc and dmc. this storm presumably lit fire wawa 41/95. the ignition analysis was carried out as an anecdotal discussion of conditions at the time of ignition of the two fires that were ignited in the three years under study (1994-1996). the conditions surrounding these two ignitions, both in june of 1995, were consistent with other descriptions of fire weather in the pukaskwa area. warm and dry weather with a high barometric pressure preceded a lightning storm. the geographic analysis suggested that there might be relatively little reason to modify fire policy based on lightning density within the park. the weather analysis suggested that it is rare to have the right combination of events that might ignite fires- dry fuel and lightning.