This research evaluated key drivers of satisfaction with cellular network performance and quantified the relative order of importance for each of the drivers. The study also validated an existing survey instrument, and explored an expanded conceptual framework that draws on service and product quality literature to suggest additional issues and attributes to investigate in future efforts to understand and model perception of cellular network quality. Additional attributes explored were expectation, level of use, other service features, and personality.;The body of published research is generally based on tracking studies that utilize univariate data analysis such as top-box and proportions (Power and Associates, 2003. The literature review in this study reaffirmed several key network attributes commonly surveyed in satisfaction surveys (network availability, coverage, drop calls, and call quality), and also determined the relative impact of each of the variables on satisfaction with network performance.;With respect to descriptive statistics, there are lots more males than female, and there are considerable differences in size and number of account sizes and types. However, descriptive results showed call quality with highest satisfaction level, followed by network availability, drop calls and coverage with mean satisfaction values of 3.68, 3.38, 3.26, and 3.02 respectively. Box-Cox transformation of the dependent variable improved the linearity of the regression model by a modest value of .6% in total variation. Multiple regression analysis was applied to examine the effects of each independent variable on network satisfaction and rank relative order of importance. Together, the independent variables explained approximately 37% of the variation in the dependent variable. With outliers removed, the model explained nearly 45.3% of total variation in network satisfaction. "Network availability" emerged as the most highly correlated predictor to network satisfaction, followed by "coverage" and "call quality" with regression beta values of .435, .174 and .125 respectively.;Lastly, the normality assumption of regression was met in which the residuals were normally distributed and constant variance (homoscedastic) over sets of values of the independent variables. However, studentized vs. predicted Y plot revealed a slight deviation from linearity of datapoints. Multicollinearity was also assessed and was not a problem.