Longitudinal gait measures may be used to provide baseline data for intervention studies. This has not previously been reported in people with Alzheimer's Disease. In this study measures of walking and their variability were recorded for 19 people with Alzheimer's Disease on two occasions 1 year apart. Matched controls were measured once. Variability was calculated using the coefficient of variation (CV). Effect size was calculated using Cohen's d. Gait was slower and more variable in the Alzheimer's Disease group compared to controls. Over 1 year there was a decrease in velocity (initial=103.9cm/s, follow-up=95.1cm/s; p<0.05, d=0.4) and stride length (initial=119.6cm, follow-up=112.5cm; p<0.05, d=0.34) and an increase in double support (initial=24.2%, follow-up=30.1%; p<0.05, d=0.99) and stride length variability (initial CV=3.5%, follow-up CV=4.6%; p<0.05, d=0.65). These changes occurred in mild as well as more severe Alzheimer's Disease. Future research should focus on reducing this decline early in the course of the disease in order to maintain physical independence for as long as possible.