The tropical cyclone forecasting skill level of the Joint Typhoon Warning Center (JTWC) has reached a plateau and has shown little or no lasting improvement during the last 10 years. Because JTWC relies on an array of statistical and climatological forecast aids for objective guidance, it seems reasonable to conclude that the skill of these aids has also reached an upper limit. We compare the forecast skill of the Navy's new “two-way interactive” nested model with the official forecast of JTWC for a large number (220) of tropical storm and typhoon cases over the period 1975–80. Even with this relatively simple model, significant potential for improved operational forecast accuracy is apparent, especially in the longer range (48–72 h) forecasts. Skill at still longer periods is also demonstrated in the results of 50 five-day forecasts. We conclude that dynamic models offer great promise for more, accurate forecasts.