In real-life applications of geometric programming, some of the coefficients and/or exponents need not be precisely known. We indicate which of general stochastic programming techniques and under which structural and distributional assumptions do not destroy the favorable structure of geometric programs.