We derive stellar ages, from evolutionary tracks, and metallicities, from Stroemgren photometry, for a sample of 5828 dwarf and sub-dwarf stars from the Hipparcos (Cat. <I/239>) Catalogue. This stellar disk sample is used to investigate the age-metallicity diagram in the solar neighbourhood. Such diagrams are often used to derive a so called age-metallicity relation. Because of the size of our sample, we are able to quantify the impact on such diagrams, and derived relations, due to different selection effects. Some of these effects are of a more subtle sort, giving rise to erroneous conclusions. In particular we show that [1] the age-metallicity diagram is well populated at all ages and especially that old, metal-rich stars do exist, [2] the scatter in metallicity at any given age is larger than the observational errors, [3] the exclusion of cooler dwarf stars from an age-metallicity sample preferentially excludes old, metal-rich stars, depleting the upper right-hand corner of the age-metallicity diagram, [4] the distance dependence found in the Edvardsson et al. (1993, Cat. <J/A+A/275/101>) sample by Garnett & Kobulnicky (2000ApJ...532.1192G) is an expected artifact due to the construction of the original sample. We conclude that, although some of it can be attributed to stellar migration in the galactic disk, a large part of the observed scatter is intrinsic to the formation processes of stars.