Stretched belemnites, and other fragmented rigid-brittle inclusions, offer great potential for estimating true extensions in deformed rocks. In the standard (Ramsay) method of analysis the elongation is calculated as 1 + e = L/L o where L is the final end-to-end length of the fragmented belemnite and L o is the original length (sum of the individual fragment lengths). This method always underestimates the true extension, usually by a substantial amount. An apparently trivial modification of the standard method, as used by Hossain, gives much improved extension estimates without any additional cost in time or effort. The strain-reversal method proposed by Ferguson is the most accurate method currently available but involves more laborious data collection, and subsequent computer processing. The three methods were applied to 120 stretched belemnites in the external French Alps. The different extension values calculated by the three methods arise because Ferguson's method is sensitive to the entire fracture-and-separation history, Hossain's method is sensitive to only part of that history, and Ramsay's method is history insensitive. If strain history is of prime interest Ferguson's method should be used, but the ease, speed, and relatively good performance of Hossain's method will make it attractive to many field geologists.