A fiber optic conjugate stress sensor (FOCSS) is explored for monitoring fatigue in aluminum cantilever beams. It is known that as damage accumulates due to fatigue, microstructural changes result in an altered modulus; so in this work, fatigue is continuously monitored by measuring the evolution of the effective modulus of the host. Conjugate stress sensing extracts host modulus from two extensometers of different stiffnesses (termed the stiff and compliant sensors) by using a known transfer function from sensor strains to host modulus. The sensing concept is explained and the transfer function is determined from a lumped-mass spring model. Finite element modeling is carried out, which reveals the host modulus assessed by the FOCSS is more sensitive to fatigue damage of a structure than its natural frequency (another commonly used macroscale measure of fatigue). The FOCSS capabilities are next verified in aluminum beam fatigue experiments. The FOCSS estimated modulus is seen to decrease with accumulation of fatigue damage. However, fatigue of the glue used to attach the sensors to the specimen is found to obscure the sensor’s response to host fatigue, thus presenting an important source of experimental error. Future work aims to deconvolve FOCSS sensitivity to glue fatigue from the response to host fatigue. Our finite element simulation results indicate that the FOCSS can significantly hasten early warning and extend the prognostic horizon, which can greatly reduce logistical burden and increase consumer confidence.