We wished to determine if an electrode polarity effect existed in vacuum breakdown conditioning. To investigate this conditioning effect, we used a three‐electrode experimental tube constructed in such a manner that one electrode (1) could be made to oppose either of the other two (2 or 3) without destroying the vacuum. High‐current arcs were drawn between both possible electrode pairs to bring them to a reproducible state of apparent deconditioning. These pairs (1 vs 2 and 1 vs 3) were then, in turn, conditioned by a series of high‐voltage pulses at a gap of 2.3 mm. The following conclusions were drawn: (1) within experimental error, all conditioning took place at the pulse cathode, (2) when the arc and pulse polarities were similar, the subsequent pulse breakdown voltage was lower than when the polarities were opposed, and (3) conditioning was not due to the removal of a gas layer from the electrode, but, rather, to removal of cathode deformations. Predictions concerning the polarity effect in vacuum breakdown conditioning were drawn from the major theories of vacuum breakdown, and compared with our experimental results. We found that our results agreed with the predictions of some vacuum breakdown theories, flatly contradicted the predictions of others, and tended to disagree with the predictions of a third group of theories.