Gravitational lensing causes background galaxy images to become aligned, and the statistical characteristics of the image alignments can then be used to constrain the power spectrum of mass fluctuations. Analyses of gravitational lensing assume that intrinsic galaxy alignments are negligible, but if this assumption does not hold, then the interpretation of image alignments will be in error. As gravitational lensing experiments become more ambitious and seek very low-level alignments arising from lensing by large-scale structure, it becomes more important to estimate the level of intrinsic alignment in the galaxy population. In this article, I review the cluster of independent theoretical studies of this issue, as well as the current observational status. Theoretically, the calculation of intrinsic alignments is by no means straightforward, but some consensus has emerged from the existing works, despite each making very different assumptions. This consensus is that a) intrinsic alignments are a small but non-negligible (< 10%) contaminant of the lensing ellipticity correlation function, for samples with a median redshift z = 1; b) intrinsic alignments dominate the signal for low-redshift samples (z = 0.1), as expected in the SuperCOSMOS lensing survey and the Sloan Digital Sky Survey

8 pages. Invited talk at Yale Workshop on `The Shapes of Galaxies and their halos', May 2001