A mathematical model was developed to predict the content of undegradable neutral detergent insoluble protein (UNDIP) from chemical characteristics of tropical forages. This study was based on a biological limitation of a previous model, which restricts the UNDIP estimates to values equals or higher than 1.34% of dry matter. The databank was formed by 540 samples of tropical forages used in cattle feeding (fresh forage and hay). The ratio of UNDIP on neutral detergent insoluble protein (NDIP) was chosen as response variable and the acid detergent insoluble protein (ADIP) as independent variable. The mathematical model was found to be exponential, assuming the formula: UNDIP = NDIP x e-(0,818+0.16764DIP) , in which all values are expressed on dry matter basis. It was observed that biological limitation of the previous model was eliminated, even though a low statistical improvement was obtained. The prediction of a biological parameter (UNDIP) from a chemical characteristic (ADIP) still have some restrictions and the estimates should be applied with caution in some situations. The main application of the model described above is estimate UNDIP contents when biological methods are not available.
acid detergent insoluble protein; digestible protein; rumen degradation; protein degradation