Both consumer and descriptive analyses were used to study the shelf-life of packaged, roasted and ground coffee in order to evaluate the utility of using sensory data in a shelf life model, and to examine the ability of consumers to detect a difference between treated and control samples. Two types of coffee, which differed in bean roast, blend and origin, were evaluated and were referred to as moderate roast and high roast. Samples were packaged under <2%, 9%, or 21% oxygen in airtight laminated foil packages. Over the course of 84 days, the samples were stored in 20��C, 30��C, or 40��C conditions and samples for all treatments were removed throughout the storage. The control samples were packaged at <2% oxygen and stored at -35��C for the duration of the study. Descriptive results illustrated a time trend in the moderate roast coffee along Factor 1, characterized by changes in the attributes of paper aroma, paper flavor, paper aftertaste, wood flavor, body, and oily mouthfeel. This trend resulted in significant differences between the control and the 40��C high oxygen 7 day stored samples in paper aroma (p=0.000) and paper flavor (p=0.000), as well as significant differences between the control and the 40��C high oxygen 49 day stored samples in paper aroma (p=0.002), paper flavor (p=0.000), paper aftertaste (p=0.000), body (p=0.000), and wood flavor (p=0.002). Factor I data from the 40��C high oxygen, moderate roast coffee were utilized in an Arrhenius shelf-life model. A second order model was fit to the data, indicating curvature in relationship for the rate of change over temperature for Factor 1 attributes. The model successfully generated a predictive curve for storage at 40��C. Consumers evaluated 30��C stored samples packaged under high oxygen and stored for either 4 or 7 days, and packaged under low oxygen and stored for 42 and 84 days. Consumer results indicated that it was difficult for consumers to detect differences between the moderate roast coffees. Consumers were readily able to tell significant differences between the high roast coffee treatments and the control sample (p<0.05). / Graduation date: 2000