This study has investigated the influence of varying the ratio of paraffin, naphthene and aromatic (P/N/A) components in narrow, wide and real petroleum fractions on the parachors of the fractions. Surface tension and density measurements have been made on synthetic hydrocarbon mixtures simulating real SC8, SC10, SC13 groups, their mixtures and a kerosene sample covering the range SC4-SC14 in order to determine their experimental parachors. Two methods to compute the parachors of fractions are proposed based on the equation of Broseta, the correlations of Riazi and Al-Sahhaf and the pseudocompound approach of Daubert. The first, denoted method I, is given by P-a (cut) = [0.85 - 0.19 omega (cut)]T-c(12/11) (cut)/P-c(9/11) (cut) and the second, denoted method II, by P-a(cut) = z(P)P(a)(P) + z(A)P(a)(A) + z(N)P(a)(N) where P-a is the parachor of the cut, omega the acentric factor, P, N and A the P/N/A components, z(P/N/A) the mole fractions of the P/N/A components and the other terms have their usual meaning. Both of these methods were found to compute the parachors of mixtures representing the SC8, SC10 and SC13 groups, mixtures of these groups and the kerosene sample with an absolute deviation of 2%. Comparison of the experimental parachors with calculated ones show that the parachor of a petroleum fraction depends not only on its molecular weight but also on the ratio of the P/N/A components. It was also found that the parachor of each of the families of P, N and A components could be expressed as a linear function of molecular weight leading to the following expression for a mixture containing P, N and A components P-a(cut) = (z(P)A(P) + z(N)A(N) + z(A)A(A))MW + (z(P)B(P) + z(N)B(N) + z(A)B(A)) where the A and B terms were those found in the linear regression fit for each family. The approach proposed gives quantitatively better results than methods such as that of Fawcett, where the P/N/A distribution is not taken into account in the computation of petroleum fraction parachors.