As the most important and fragile part of fiber optic gyroscopes (FOG), the reliabiliey of superluminescent diodes (SLD) determines the operational reliability and maintenance expenses of FOGs. Reliability of SLDs is studied based on degradation data for their long life characteristics. Based on analysis of degradation mechanisms of SLDs under environmental stress firstly, normal-poisson hybrid stochastic process model is presented as their life distribution model whose parameters can be estimated from the degradation measures of the SLDs. The proposed method can obtain SLDs′ reliability without failure data, therefore overcomes the drawbacks of traditional time-to-failure analysis method which requires failure data. This estimation method shows its significance in test costs and time.