A method of estimating bias errors for an air vehicle in flight under high spin conditions is provided. The method comprises collecting a first set of data samples at a first time and collecting a second set of data samples at a second time. The first time and the second time occur after deployment of the air vehicle and before canard shock. The first and second sets of data samples comprise roll rate data from at least a first gyroscope and acceleration data from at least a first accelerometer. At least a first estimated accelerometer bias from the first and second sets of data samples is calculated.