In order to check the evolutionary status and theoretical models of eclipsing binaries of Algol type, a reliable determination of their absolute dimensions is needed. In this communication, we compare the most commonly used methods to derive absolute parametors in single-lined aclipsing binaries. Let us first assume that the mass function, f(m), is known from the analysis of the radial velocity curve while the relative radii and orbital inclination are derived from the light curve. The determination of absolute parameters is then equivalent to the obtention of the mass ratio, q = m2/m1. The following methods are available to estimate q from observed parameters — over-all errors being estimated for observational uncertainties of the order of 5 % in relative radii and temperatures and 15 % in f(m) —:1.qs: It is assumed that the primary component follows the mass-luminosity relation for main-sequence stars. This procedure provides qs with an uncertainty of about 10 %.2.qsD: It is assumed that the secondary component fills its Roche lobe. Errors of at least 15 to 20 % are expected from this procedure mainly due to its high sensitivity to small variations in the observed value of r2, particularly if r2 > 0.2.Both methods can be used together when f(m) is doubtful, or completely unknown, but errors can not be expected to be better than in case 2.3.qn: It is assumed that the primary component rotates synchronously in a circular orbit. This assumption is difficult to adopt due to the expected transfer of angular momentum through mass transfer and the value of qn is estimated with about 20 to 30 % error.4.qED: It is assumed that the primary component, is well reproduced by standard evolutionary models within the main sequence. Adopting a grid of models for a given chemical composition, an iterative procedure in the log Ti,.–log g plane permits the determination of m1 and thus q. This method is equivalent to (1) but avoids errors due to evolution from the ZAMS to the TAMS, not taken into account in the previous method, and allows to reach a higher accuracy, around 5 %, except for those primary stars located around the TAMS, where the determination of mi is not unique.