Tensile data of various rotor and ring spun knitting yarns with different finishing treatments are analyzed in order to produce a correlation with yarn performance and production parameters. Among the fitted theoretical distributions, the Weibull and the Normal distribution function are closest to the experimental data distribution. The fit of both distributions deteriorates as the data increases. The data are also analyzed statistically with the "Moving Average" test and a Spectral Analysis. From the wavelengths of obvious periodic errors in the yarns it can be seen that these were produced mainly in the last production passages before spinning. Furthermore, a clear trend corresponding to a periodicity with the huge wavelength of a whole cone can be revealed for the ring spun yarns. The process parameter "spinning system" can be predicted clearly with a Neural Back-propagation Network. Knitting performance does not show a good correlation. The correlation results achieved with the Neural Network confirm that the Weibull distribution gives even a slightly better description of the experimental data than the Normal distribution.