Modeling the performance of high-pressure, high-temperature (HPHT) natural gas reservoirs requires the understanding of gas behavior at such conditions. In particular, gas viscosity is an important fluid property that directly affects fluid flow through porous media and along production flowlines. Accurate measurements of gas viscosity at HPHT conditions are both extremely difficult and expensive. Unfortunately, the correlations available today do not have a sufficiently broad range of applicability in terms of pressure and temperature since no measured gas viscosities at HPHT are currently available. Thus the correlation accuracy may be doubtful for the prediction of gas viscosity at HPHT conditions. An oscillating-piston viscometer was used to measure the viscosity of mixtures of nitrogen and methane, and mixtures of CO2 and methane at a pressure range of 5,000 to 25,000 psi, and a temperature range of 100 to 360 degrees F. The viscosity of mixtures of nitrogen and methane, and mixtures of CO2 and methane measured to take into account of the fact that the concentration of non-hydrocarbons increase significantly in HPHT reservoir. The recorded measured data were then used to evaluate the reliability of the most commonly used correlations in the petroleum industry. Measured gas viscosity data at HPHT conditions suggest that the most common gas viscosity correlations return up to 9% relative error in gas recovery factor, which translates into a significant error in estimating the ultimate recovery for large HPHT natural gas reservoirs. Thus, the current gas viscosity correlations need to be adjusted to estimate gas viscosity at HPHT conditions. New gas viscosity correlations constructed for HPHT conditions developed based upon our experiment data provide more confidence on gas viscosity. A rolling ball viscometer was also used to assess its capability to measure gas viscosity. Using gas instead of liquid to calibrate a rolling ball viscometer over the entire pressure and temperature range of interest appears to be satisfactory. Optimizing tube inclination angle and ball/tube diameter ratio prevents turbulent flow effects around the ball, thus enhancing the accuracy of the measurement. The proposed calibration method was then verified with pure CO2 at a pressure range of 4,000 to 8,000 psi, and a temperature range of 98 to 240 degrees F. Consequently, rolling ball viscometer was introduced as a good candidate to measure the gas viscosity; however it has not been tested at HPHT conditions yet.