The problem of quantifying uncertainty in hydrologic model output is examined, and a review of previous work on the topic is given. The study focuses on the estimation of error bounds for simulated hydrographs, assuming the model to be perfect and accounting only for output uncertainty caused by uncertainty in parameter values. First-order uncertainty analysis is applied to the Stanford Watershed Model to calculate the effects of parameter uncertainty as it propagates through the model. The first-order approximations to the mean and standard deviation of model output are compared to estimates of the same quantities, using Monte Carlo analysis. This was done to identify the degree of parameter error for which the first-order approximations are valid. First-order analysis was found to give good approximations if the coefficients of variation (standard deviation/mean) on sensitive parameters were less than about 0.25. First-order analysis allows a large reduction in computation over Monte Carlo methods for estimating error bounds on simulated hydrographs.