A vibrating-wire viscometer of very high precision was used to measure the viscosity of methane and of two natural gases. The experimental data were, in general, taken at temperatures of 260, 280, 300, and 320 K and at pressures up to 20 MPa, and additionally in the case of methane at temperatures of 340 and 360 K and at pressures up to 29 MPa. The estimated uncertainty is ±0.3 and ±0.5% for methane and the natural gases, respectively. The new experimental data for methane were used together with zero-density or low-density viscosity values from this study and from the literature to develop a viscosity equation for natural gas composed of two contributions. The mixing rule of Wilke [J. Chem. Phys. 18: 517 1950] was applied for the zero-density viscosity part which is based on zero-density correlations for twelve components (methane, nitrogen, carbon dioxide, ethane, propane, n-and isobutane, n-and isopentane, n-hexane, n-heptane, and n-octane) and agrees with the values derived from experiment within ±0.3%. The density dependence of the residual viscosity part was correlated with methane data only, neglecting any temperature dependence, whereas the composition dependence is characterized by a pseudo-critical viscosity value. For methane the agreement between the correlated and experimental data is within ±0.5%. The values predicted with the correlation and the experimental data agree within ±1% for both the high calorific, H, natural gas and the low calorific, L, natural gas.