An important factor in predictability, or the outcome of numerical weather forecasts, is the impact of observation data. Here, a procedure for assessing the impact of observations on short-range forecast error in numerical weather prediction is described, which uses the adjoint of a global forecast model and the adjoint of a 3-d Variational data assimilation procedure. The method is computationally inexpensive and allows observation impact to be partitioned for any set or subset of observations, by instrument type, observed variable, geographic region, vertical level or other category. Results will be shown to illustrate the impact of various types of satellite and in-situ observation data on forecast error in the global domain.