Starting from the assumption that "conditioning" is necessary to trigger a breakdown and that its duration can be seen as the delay time before breakdown, a new approach to the problem of the dielectric stress of solid insulators is proposed. It is founded on the relationship existing between the delay time which is determined statistically, and the amplitude of the electrical field. The experiments deal with a few hundred nm thick layers of silicon oxide, alumina, and polystyrene, coated with self-healing electrodes to which successive voltages of different amplitudes are applied. The delay time is determined either by τ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">I</sub> , i.e. the lag time between two consecutive breakdowns, or by t <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">I<.sub>, time during which the material has been under stress before the ith breakdown. The variations, as a function of field G, of the mean value τ<sub>m</sub> of τ<sub>I</sub>, or of tτ inferred from Weibull statistical model applied to ti, yield an asymptotic value G<sub>c</sub> for which the "conditioning" duration tends to infinity. By identifying the curve τ<sub>m</sub>(G) with an expression such as τ<sub>m</sub> = a/(G-G<sub>c</sub>)<sup>α</sup>, the three constants a, α and G<sub>c</sub> are calculated, the latter representing a "specific breakdown field"; t<sub>w</sub>(G) may give a similar definition. An example of the use of such a criterion is its variation as function of the sample thickness.</sub>