The viscosity of pure hydrocarbons was correlated using a simple function based on the fluid density. The correlation has three adjustable parameters, a compressed state density, rho(s)degrees, an empirical parameter, c(2), that scales the viscosity response to fluid expansion, and another empirical parameter, c(3), used to tune at pressures above 10 MPa. The inputs to the correlation are the fluid density, pressure, and low-pressure gas viscosity. The correlation fit experimental viscosities for 39 pure hydrocarbons including n-alkanes, branched alkanes, alkenes, cyclics, and aromatics, within experimental error over a broad range of temperatures and pressures. Heavy hydrocarbons such as mineral oils were also fit with an AAPRD of 2.7%. Binary mixture viscosities were predicted to within 10% using simple volumetric mixing rules. The method provides a single framework for liquid and vapor phases, is simple to implement, and is very fast computation ally, making it ideal for incorporation into process and reservoir simulators.