Hot plate methods under steady state conditions are based on a 1D model to estimate the thermal conductivity, using measurements of the temperatures T0 and T1 of the two sides of the sample and of the heat flux crossing it. To be consistent with the hypothesis of the 1D heatflux, either a hot plate guarded apparatus is used, or the temperature is measured at the centre of the sample. On one hand the latter method can be used only if the ratio thickness/width of the sample is sufficiently low and on the other hand the guarded hot plate method requires large width samples (typical cross section of 0.6 × 0.6 m2). That is why both methods cannot be used for low width samples. The method presented in this paper is based on an optimal choice of the temperatures T0 and T1 compared to the ambient temperature Ta, enabling the estimation of the thermal conductivity with a centered hot plate method, by applying the 1D heat flux model. It will be shown that these optimal values do not depend on the size or on the thermal conductivity of samples (in the range 0.015–0.2 W m−1 K−1), but only on Ta. The experimental results obtained validate the method for several reference samples for values of the ratio thickness/width up to 0.3, thus enabling the measurement of the thermal conductivityof samples having a small cross-section, down to 0.045 × 0.045 m2.