The Variable Stripe Length (VSL) method is a very popular tool to measure the optical gain in thin film active devices. However, over the last decade experimental and theoretical evidence has been reported that cast doubt upon its reliability and that seriously discourages its application. Continuing in the path of highlighting the uncertainties associated with this method, this Letter soundly demonstrates that the particular choice of stripe lengths in the VSL measurements profoundly influences the optical gains retrieved by this method. Thus, a single set of VSL data may render gain values that differ by tens of cm^-1. The observed gain variability is ascribed to a combination of unavoidable experimental noise and incorrect assumptions in the analytical treatment (small-signal approximation).