A sensing system has been under development for measuring cotton fiber quality as cotton is harvested in the field, enabling cotton fiber quality maps to be generated with data from the sensor and a GPS receiver. In developing the sensor, reflectance spectra of cotton fiber samples having different fiber quality levels have been measured with a high-resolution spectrophotometer. It was observed, consistent with some literature on the subject, that cotton fiber with lower micronaire values reflects more light energy at most wavelengths from 900 to 2500 nm. The relationship between micronaire and spectral reflectance was evaluated with simple and multiple linear regression models, and also wavelet analysis. These models indicate that the micronaire has a strong correlation with the reflectance values in six different wavebands. Micronaire can be predicted very well even with a model involving only two wavebands as independent variables. Based on the characteristics of the cotton fiber reflectance spectrum and the analysis model developed, a cotton fiber quality sensor was designed. The sensor consists of a near-infrared camera and a data collecting and processing system with an on-line sampling device. The camera's image is a measure of fiber reflectance at selected wavebands. The reflectance data are collected and processed with a predetermined algorithm to determine fiber properties. The sampling device is able to automatically take cotton samples from a harvester and prepare the samples in a consistent manner for reflectance measurement.