Under the assumption that a geometric process at the start point follows a lognormal distribution, the statistical inference problem for the geometric process is studied. The distribution parameters have been estimated with either the maximum likelihood method or a modified moment method. Then some suggestions are made based on the theoretical results, the simulation experiments, and the real data analysis.