Over the past four decades, tropical cyclone models have evolved from their inception from barotropic and statistical models which were primarily concerned with simulating storm tracks. It was during the 70's that specialized 3-D models were developed specifically for simulating tropical storms. As contrasted with models to simulate extra-tropical system and the global circulations, these TC models focused on one system, were locally forced and dealt with sparse data over the tropical oceans and spanned important scales from cumulus through the synoptic. Because of computation limitations (memory and speed), early 3-D models (Anthes, Kurihara, and Jones in the 1970's ) were confined to research studies since computations took longer than real time even on state of art mainframe computers. Since then gradual transition has occurred from modeling essential axi-symmetric structure to more complete simulation of 3-D spiral bands, asymmetric outflow, and eye structure in the more advanced research models developed in the community model (e.g. MM5 and WRF) framework. Early 3-D operational models were the MFM (Hovermale) and QLM (Mathur). The GFDL model with its nesting design proved to be a desirable candidate for transition to operations. It demonstrated its worth as a research model with many years of development before becoming operational in 1995. Coincident with the four decade upgrades of regional models, have been the marked improvement of global forecast systems. Significant upgrades occurred in the global forecasts systems of NCEP, the Navy and ECMWF in data assimilation, physical packages, and dynamics. Today, operational global models have resolutions that compare favorably to those of regional hurricane models a decade earlier. Over the past two decades NHC three day track errors have plummeted from ~300 nm to ~150 nm due in part to numerical guidance.