Early abundance measurements established that stars known to host giant planets are metal rich compared to the Sun. More extensive abundance measurements then showed that giant planet hosts are metal rich compared to the parent sample in planet searches. Stars spanning a range of convection zone depths all show the same metallicity effect, ruling out significant abundance enhancements due to selective accretion. Most known planets migrated inwards from the snow line, but subsamples closer to and further from the star have similar iron abundances, so the stopping point of migration does not depend on metallicity. Stars recently discovered to host Neptune mass planets may be metal poor compared to the Sun, particularly if one focusses on stars that do not also host higher mass planets. This would be consistent with core-accretion models of planet formation. Before drawing physical conclusions, it will be necessary to check for metallicity bias in the subsample of stars around which Neptune mass planets could have been found. M dwarf abundances are currently too uncertain to relate planet frequency and host star metallicity, due mainly to missing or incorrect molecular line data.