Maps of uncertainty and parametric sensitivity in ground motion have been produced for three southern California counties using a Monte Carlo approach. The uncertainty in hazard is estimated by random sampling a logic tree that has the uncertainty in most parameters described by a normal distribution. The nine parameters varied in the uncertainty analysis are those related to the estimation of recurrence rates for earthquakes on faults using geologic information, the ground-motion attenuation relations, the relations available for estimating the magnitude-frequency distribution of earthquakes on faults, and the inclusion or exclusion of blind thrusts in the Los Angeles area. Besides simultaneously varying all nine parameters for a Monte Carlo series of runs to obtain the overall uncertainty in ground motion for a fixed annual probability, each of the nine parameters were varied separately while the others were held fixed to determine the sensitivity of each parameter on this uncertainty. The total uncertainty in shaking levels corresponding to a fixed annual probability is up to ±50% at the 95% confidence level. The parameters that most influence the uncertainty in the hazard estimates from geologic information are (1) maximum magnitude, (2) choice of attenuation relations, (3) the magnitude-frequency distribution, and (4) slip rate.