Linear, quadratic, cubic, compound, exponential and logistic regression models were used to evaluate the temperature and pH dynamics of a solid-state fermented (SSF) cocoa shell (Theobroma cacao L.) treated with natural yogurt for use in animals. For this purpose, temperature and pH were checked in silage samples of 0, 1, 4, 8, 15 and 30 days. For the processing of the data, the statistical package SPSS version 22 was used. The behavior of the two variables were adjusted to a cubic model: temperature (R² = 0.88; Y = 23.8145630348868 + 0.188*X - 0.022*X² + 0.00054*X³) and pH (R² = 0.95; Y = 5.543420742606556 - 0.3620949736293026*X + 0.02686662258086579*X² - 0.0005405196051196668*X³). In conclusion, the temperature and pH behavior were adjusted to a non-linear cubic model, and were kept within the optimal ranges for the production of SSF of optimum quality for use in animals.