Method for optimizing the development of an underground reservoir, wherein a geological model is updated using dynamic data and well tests.A reservoir model is constructed by performing a geological model scale change. Dynamic data are simulated from this reservoir model. Influence zones are identified within the geological model where the well tests induce a pressure variation during well testing. Well tests are then simulated for each influence zone. An objective function measuring the difference between the simulated data and the measured data is calculated. The geological model is then modified so as to reduce to the maximum the objective function using a geostatistical parametrizing technique. Finally, development of the underground reservoir is optimized by evaluating, by means of a flow simulator, the reservoir production for various production schemes.Application: notably oil reservoir development.