F-15 taxi tests and analyses were performed to evaluate the effects of aircraft design changes on nose landing gear shimmy. Preliminary analytical studies indicated that these changes would have an adverse effect on shimmy speed. This was of particular concern because limit cycle shimmy had been experienced on the baseline gear for cases with out-of-tolerance strut torsional freeplay. The tradeoffs considered in the choice of a taxi test over a laboratory dynamometer test are presented. Operational aspects of the taxi test are discussed. Several instances of limit cycle shimmy were encountered during testing and results indicate that shimmy speed is a function of strut torsional freeplay. A description of the math model used in the non-linear analyses is provided. Analytical results are presented in terms of shimmy speed versus strut torsional freeplay. These results confirm the limit cycle nature of the shimmy phenomenon and correlate well with the taxi test results. Additional analyses are presented indicating the sensitivity of the shimmy to changes in tire parameter values and strut frictional coefficients. Assumptions used in the development of an equivalent linear math model are given.