Sixteen precision gravity surveys carried out since 1979 provide an additional constraint for numerical simulation of the Bulalo system. The gravity surveys were all run using the same LaCoste and Romberg gravity meter (D-33). Leveling surveys collected in conjunction with the gravity surveys used Wild N3 spirit levels. The leveling surveys were used to correct the gravity data for elevation changes. Since production began in 1979, the Bulalo field has experienced a gravity change of up to -600 microgal. The decline in gravity is interpreted to result from density changes in the reservoir, primarily due to the development of a two-phase zone where liquid saturation is lower in both the rock matrix and its fracture network. An accurate match of the trends and magnitudes of observed and simulated gravity changes provides a constraint for the mass balance of the numerical model beyond the matching of pressure and enthalpy evolution. Because of internal flow effects in most of the wells, pressure and temperature measurements do not necessarily represent static formation pressure and temperatures. Gravity data, therefore, became particularly important for calibrating the net depletion of the numerical model and for constraining reservoir mass replacement from reinjection and natural inflow from aquifers outside the production zone. Early comparisons of the observed and simulated gravity showed a too rapid gravity decrease prior to 1992, and a too slow gravity decrease after that time. Subsequent changes made to the numerical model, including modifying permeability structure and in-flow direction, resulted in an improved match to the gravity and available pressure data. The result has been a numerical model that accurately matches not only the quantity of net mass withdrawal, but also the location of the mass withdrawal including recharge from outside the producing reservoir. (C) 2003 CNR. Published by Elsevier Ltd. All rights reserved.