Global climate models are mathematical models which mimic the fluid dynamics and physics of the Earth's atmosphere, oceans, and land surface. Scientists have developed these models to a state of sophistication such that they reproduce many observed features of the current climate, and the most powerful supercomputers are required to obtain a solution. They are able to simulate the large scale features of the climate, and they produce a facsimile of the observed annual cycle and other variations of Earth's climate. Nevertheless, the models used to study the greenhouse warming problem are not perfect. Global climate models used around the world do not completely agree on all aspects of climate variability or sensitivity, and they all have serious deficiencies. For example, the atmospheric component of the best available climate models have seasonal errors of hemispheric scale surface air temperature of 1.5° - 2.5° C on average, and they misrepresent hemispheric precipitation by 20-25% (root mean squared errors). Furthermore, model biases and inter-model differences are too large to make possible any inferences about regional climate changes.