This paper is concerned with sensitivity analysis for the distributed parameter estimation problem arising in the modeling process for fluid flow in underground porous media. An efficient algorithm was constructed using variational calculus techniques for the evaluation of sensitivity gradient curves which describe variations of system outputs resulting from variations of a spatially-vary-ing parameter in nonlinear partial differential equations. For a test problem of estimating transmissivity of a one-dimensional ideal gas reservoir, sensitivity behavior was analyzed under various reservoir conditions, and the results were applied to devising a parameter discretization scheme which will yield improved parameter estimates.