Cross‐sensor beamforming offers the potential of providing good side‐lobe suppression for sparse acoustic arrays or filled arrays which develop faulty elements during exercises at sea. To determine the sensitivity of this technique to random errors in amplitude and phase we computed the cross‐sensor beam pattern with each element in the array subject to a random amplitude and phase error. For a line array with 32 equally spaced elements, the beam pattern from the cross‐sensor method was nearly identical to that of the conventional beamformer with both subject to the same set of random errors. Repeating the same computation with four inoperative elements showed that the cross‐sensor method maintained good side‐lobe suppression while the side lobes of the conventional beamformer were elevated corresponding to the amplitude weights of the faulty elements. The mean beam pattern for the cross‐sensor method was computed analytically assuming the errors were random, independent, and uniformly distributed. The result can be expressed as the sum of the error‐free pattern and an additional angle‐independent term due to the random errors. Comparison of this expression with the corresponding one for the conventional beamformer reveals that for a filled array the two results are identical. [Work supported by NAVMAT.]