Two methods of estimating expiration date have been compared. For a drug whose potency is expected to decrease with time, method 1 fits a linear regression line to the stability data, calculates a 95% lower confidence limit of the true degradation line, and estimates the expiration date by comparing the confidence limit with the lower specification limit. Method 2 calculates a 95% lower confidence limitb1of the slope of the true degradation line, constructs a new line y” usingb1as its slope, and finally estimates the expiration date by comparing y” with the lower specification limit. In this work we show that in most practical situations, method 1 estimates greater expiration date than method 2.