We have measured the temperature dependence of the conductance in long V-groove quantum wires (QWRs) fabricated in GaAs/AlGaAs heterostructures. Our data is consistent with recent theories developed within the framework of the Luttinger liquid model, in the limit of weakly disordered wires. We show that for the relatively low level of disorder in our QWRs, the value of the interaction parameter g ∼ = 0.66, which is the expected value for GaAs. However, samples with a higher level of disorder show conductance with stronger temperature dependence, which does not allow their treatment in the framework of perturbation theory. Fitting such data with perturbation-theory models leads inevitably to wrong (lower) values of g. The electrical conductance through noninteracting clean quantum wires (QWRs) containing a number of one-dimensional subbands is quantized in the universal unit, as observed in narrow constrictions in 2D electron gas (2DEG) systems [2,3]. For such short and clean narrow wires, the e-e interactions described by the so-called Luttinger liquid (LL) model [4] do not affect the value of the conductance, namely it is temperature and length independent as indeed was shown experimentally [2,3]. In the presence of disorder in sufficiently long QWRs, suppression of the conductance is expected at low temperatures. A number of theoretical papers addressing this issue [5,6,7,8] predict a negative correction to the conductance versus temperature G(T ), which increases with T and obeys a power law: T g−1 , where g < 1 is an interaction parameter.The validity of the implications of the LL theory has been recently demonstrated in a number of experiments [9,10]. The most evident proofs of the predictions were shown in tunnelling experiments performed in T-shaped cleaved-edged overgrown GaAs quantum wires [9] and in carbon nanotubes [10]. Earlier non-tunnelling experiments, in which suppression of conductance occurs in the linear response regime, did not unambiguously prove the validity of the theory, and the value of the g parameter could not be deduced from the experimental data [11,12,13]. Several complications are encountered in such experiments. For sufficiently disordered wires, where the correction to G(T ) is expected to be large, the value of the conductance at the plateau is not well defined due to the specific realization of the disordered potential in the wire, as was the case for the long wires of Tarucha et al. [11]. Moreover, in the intermediate regime, namely for disorder level for which the conductance plateau could be well defined but the corrections to G(T ) are already significant for a relatively narrow temperature range, g cannot be extracted by applying a perturbation theory. If however, the disorder is very weak so that the plateaus are well defined at all temperatures [11,12], the variation of its value versus temperature is so weak that the g parameter cannot be reliably determined. Therefore, if one wishes to compare G(T ) to the theory, a wire possessing just the right amount of disorder is ...