The performance accuracy of an inertial navigation system can be expressed in terms of a series of equations. Inaccuracies arise in such a system because of initial alignment errors, imperfections in the performance of the inertial instruments and limitations in the computational process. These errors can be quantified enabling a designer to estimate the performance accuracy of a proposed navigation system. The analysis can be simplified in some circumstances, such as for very short duration flight. In other cases, particularly where there is coupling between channels, a deterministic solution is not possible, and simulation is necessary to provide accurate information on performance. Inertial sensors are sensitive to various external stimuli. Fortunately these sources of error are frequently well behaved and consequently can be expressed as a deterministic equation, the coefficients of each term representing the various sensitivities to a given stimulus. Care must be taken when processing the various sensor signals in the presence of angular motion, particularly in the presence of coning and sculling motion. In these cases, the bandwidths of the sensors and the speed of the computation must be high enough to sense and record the actual motion, otherwise significant errors can arise, even if 'perfect' sensors were to be available.