This paper examines the psychometric properties of an existing measure of mental health, the GHQ‐12, as revealed in three studies involving employees in an engineering firm (n = 659), recent school‐leavers (n = 647), and unemployed men (n = 92). The measure was shown to be psychometrically sound in all cases, with a Likert scoring method providing a more acceptable distribution of scores than the more commonly deployed ‘ GHQ score ’ for use in parametric statistical analyses. Scores on GHQ‐12 were found to be much higher (indicating lower mental health) for those who were unemployed, higher for women than for men in one sample, and unrelated to age, job level and marital status.