An Elman network model was trained using Fletcher-Reeves Update training algorithm, and used to predict the tensile strength of cotton yarns. The cotton lint and yarn samples (consisting of rotor and ring spun yarns) were collected from four textile factories in Kenya. A total of 410 yarn samples were collected. These samples were divided into nine categories. Process-wise three of the yarn categories (of Ne 27, 12.5 and 7.5) were spun on the rotor spinning system, while the other six (of Ne 30, 24 and 20) were spun on the ringframe. The HVI characteristics of the cotton lint together with the fineness and twist per inch (TPI) of the yarn samples were used as inputs to the Elman network. The output of the network was set as the tensile strength of the yarns. The optimum network had 6 neurons in the hidden layer and used gradient descent with momentum as the weight/bias learning function. The Elman network predicted the yarn tensile strength of the yarn samples with a mean squared error (mse) of 0.0156 and a correlation coefficient (R-value) of 0.974.