Following the work of Peneloux et al. (1982), several attempts have been made to improve density calculations from equations of state by the use of alternative volume translation methods (Soreide, 1989; Magoulas and Tassios, 1990; Coniglio; 1993). The present work aims at improving these predictions, particularly in the case of oil and gas reservoirs (temperature up to 200 degrees C and pressures as high as 120 MPa), using the Peng-Robinson equation of state. As Volume translation methods sometimes show inconsistencies at high pressure (e.g., negative thermal expansion coefficient) we had to develop an original expression. For this purpose, we used high pressure density measurements instead of saturated liquid densities. Several pure hydrocarbons from various families were considered: C-6 to C-40 n-alkanes, cyclohexane, C-6 to C-12 monoaromatics. Within a good approximation, the volume translation c is thus shown to follow a linear dependancy with temperature and with molecular weight: c(T) = (0.023 - 0.00056 MW)T + (-34.5 + 0.4666 MW) where c is in cm(3)/mol, T in K and MW in g/mol. When this expression is used, average errors at high pressure are low (less than 3% for the C-6 to C-13 hydrocarbons investigated). For heavier hydrocarbons, its predictions are sensitive to critical properties and the following expression is recommended: C(T) = v(T-ref, P-ref) -MW/rho(ref) + (0.023 - 0.00056 MW) (T - T-ref) where v(T-ref, P-ref) is the liquid molar volume computed by the untranslated equation of state in the same conditions as those where a density measurement (rho(ref)) is available. With the exception of the near-critical region, the method is still accurate at low pressure. The method does not display the inconsistencies of previous methods, and it is also shown to be consistent with mixing rules when several pure components are lumped into a single pseudo-component. As a consequence, its application to real fluids looks promising.