In this paper, spinyhead croaker (Collichthys lucidus) fat content forecasting model using electronic nose (e-nose) and non-linear stochastic resonance (SR) has been studied. Spinyhead croaker samples are stored at 4 °C temperature. Physical/chemical indexes (firmness, total volatile basic nitrogen (TVB-N), pH, and fat content) are examined to provide quality references for e-nose. E-nose responses are treated by principal component analysis (PCA), bistable SR, and double-layered cascaded serial SR (DCSSR). SR and DCSSR SNR maximal (SNR-Max) values discriminate croakers clearly. Multi-variables regressions (MVR) are conducted between physical/chemical indexes and SR/DCSSR SNR-Max values. MVR results demonstrate that DCSSR feature values have more significant linearity relation with physical/chemical indexes. Spinyhead croaker fat content forecasting model is developed via linear fitting regression on SR SNR-Max values. Validating experimental results demonstrate that the developed model has good accuracy.