The sensitivity of energetic materials to initiating stimuli is one of the tests with binary response. Usually, there is not a single sharp boundary between energy levels causing initiation and not causing initiation. Instead, there is an interval of energies causing the initiation with certain probability, called the sensitivity curve. In the past, various methods were developed to determine the whole sensitivity curve, or its important points (e. g. Bruceton staircase, Robbins-Monroe, Langlie, Probit analysis, or Neyer'sD-optimal test, 3pod). All these methods, despite frequently used, have their limitations. We would like to introduce the new method/algorithm, called FEST (Fast and Efficient Sensitivity Testing), for the determination of a sensitivity curve. The sensitivity curve is represented by the cumulative distribution function for a lognormal distribution. The calculation of the level for the next shot is similar to Neyer's approach in the beginning of the test procedure. Later, after the overlap is reached and therefore unique maximum likelihood estimates for mu and sigma exist, the next shot level is calculated from these parameters using two user-defined constants. These constants can be used to shift the levels of testing into the area of interest of the sensitivity curve. In this article, the algorithm is introduced, its convergence to real values is supported by simple Monte Carlo simulations, and a real life example (determination of sensitivity to electrostatic discharge for a pyrotechnic mixture) is presented.