A microgrid including distributed generators can operate connected to the main electrical network or in an isolated manner, referred to as island operation. The transition between both states can occur voluntarily, but a disconnection can also happen unexpectedly. The associated transients can be harmful to the grid, and compensating actions must be triggered to avoid service interruption, preserve power quality, and minimize the possibility of faults; island detection methods are essential to this end. Such techniques typically depend on communication networks or on the introduction of minor electrical disturbances to identify and broadcast unexpected islanding events. However, local energy resources are distributed, variable, and are expected to be integrated in a plug-and-play manner; then, conventional island detection strategies can be ineffective as they rely on specific infrastructure. To overcome those problems, this work proposes a straightforward, distributed island detection technique only relying on local electrical measurements, available at the output of each generating unit. The proposed method is based on the estimated power-frequency ratio, associated with the stiffness of the grid. A “stiffness change” effectively reveals island operating conditions, discards heavy load variations, and enables independent (distributed) operation. The proposal was validated through digital simulations and an experimental test-bed. Results showed that the proposed technique can effectively detect island operation at each generating unit interacting in the microgrid. Moreover, it was about three times faster than other reported techniques.